OpenCloudOS-Kernel/kernel/locking/mutex.c

1094 lines
27 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* kernel/locking/mutex.c
*
* Mutexes: blocking mutual exclusion locks
*
* Started by Ingo Molnar:
*
* Copyright (C) 2004, 2005, 2006 Red Hat, Inc., Ingo Molnar <mingo@redhat.com>
*
* Many thanks to Arjan van de Ven, Thomas Gleixner, Steven Rostedt and
* David Howells for suggestions and improvements.
*
mutex: implement adaptive spinning Change mutex contention behaviour such that it will sometimes busy wait on acquisition - moving its behaviour closer to that of spinlocks. This concept got ported to mainline from the -rt tree, where it was originally implemented for rtmutexes by Steven Rostedt, based on work by Gregory Haskins. Testing with Ingo's test-mutex application (http://lkml.org/lkml/2006/1/8/50) gave a 345% boost for VFS scalability on my testbox: # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 296604 # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 85870 The key criteria for the busy wait is that the lock owner has to be running on a (different) cpu. The idea is that as long as the owner is running, there is a fair chance it'll release the lock soon, and thus we'll be better off spinning instead of blocking/scheduling. Since regular mutexes (as opposed to rtmutexes) do not atomically track the owner, we add the owner in a non-atomic fashion and deal with the races in the slowpath. Furthermore, to ease the testing of the performance impact of this new code, there is means to disable this behaviour runtime (without having to reboot the system), when scheduler debugging is enabled (CONFIG_SCHED_DEBUG=y), by issuing the following command: # echo NO_OWNER_SPIN > /debug/sched_features This command re-enables spinning again (this is also the default): # echo OWNER_SPIN > /debug/sched_features Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-01-12 21:01:47 +08:00
* - Adaptive spinning for mutexes by Peter Zijlstra. (Ported to mainline
* from the -rt tree, where it was originally implemented for rtmutexes
* by Steven Rostedt, based on work by Gregory Haskins, Peter Morreale
* and Sven Dietrich.
*
* Also see Documentation/locking/mutex-design.rst.
*/
#include <linux/mutex.h>
#include <linux/ww_mutex.h>
#include <linux/sched/signal.h>
#include <linux/sched/rt.h>
#include <linux/sched/wake_q.h>
#include <linux/sched/debug.h>
#include <linux/export.h>
#include <linux/spinlock.h>
#include <linux/interrupt.h>
#include <linux/debug_locks.h>
#include <linux/osq_lock.h>
#include "mutex.h"
#ifdef CONFIG_DEBUG_MUTEXES
# define MUTEX_WARN_ON(cond) DEBUG_LOCKS_WARN_ON(cond)
#else
# define MUTEX_WARN_ON(cond)
#endif
void
__mutex_init(struct mutex *lock, const char *name, struct lock_class_key *key)
{
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
atomic_long_set(&lock->owner, 0);
raw_spin_lock_init(&lock->wait_lock);
INIT_LIST_HEAD(&lock->wait_list);
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
#ifdef CONFIG_MUTEX_SPIN_ON_OWNER
osq_lock_init(&lock->osq);
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
#endif
debug_mutex_init(lock, name, key);
}
EXPORT_SYMBOL(__mutex_init);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
/*
* @owner: contains: 'struct task_struct *' to the current lock owner,
* NULL means not owned. Since task_struct pointers are aligned at
* at least L1_CACHE_BYTES, we have low bits to store extra state.
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
*
* Bit0 indicates a non-empty waiter list; unlock must issue a wakeup.
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
* Bit1 indicates unlock needs to hand the lock to the top-waiter
* Bit2 indicates handoff has been done and we're waiting for pickup.
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
*/
#define MUTEX_FLAG_WAITERS 0x01
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
#define MUTEX_FLAG_HANDOFF 0x02
#define MUTEX_FLAG_PICKUP 0x04
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
#define MUTEX_FLAGS 0x07
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
/*
* Internal helper function; C doesn't allow us to hide it :/
*
* DO NOT USE (outside of mutex code).
*/
static inline struct task_struct *__mutex_owner(struct mutex *lock)
{
return (struct task_struct *)(atomic_long_read(&lock->owner) & ~MUTEX_FLAGS);
}
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
static inline struct task_struct *__owner_task(unsigned long owner)
{
return (struct task_struct *)(owner & ~MUTEX_FLAGS);
}
bool mutex_is_locked(struct mutex *lock)
{
return __mutex_owner(lock) != NULL;
}
EXPORT_SYMBOL(mutex_is_locked);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
static inline unsigned long __owner_flags(unsigned long owner)
{
return owner & MUTEX_FLAGS;
}
static inline struct task_struct *__mutex_trylock_common(struct mutex *lock, bool handoff)
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
{
unsigned long owner, curr = (unsigned long)current;
owner = atomic_long_read(&lock->owner);
for (;;) { /* must loop, can race against a flag */
unsigned long flags = __owner_flags(owner);
unsigned long task = owner & ~MUTEX_FLAGS;
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
if (task) {
if (flags & MUTEX_FLAG_PICKUP) {
if (task != curr)
break;
flags &= ~MUTEX_FLAG_PICKUP;
} else if (handoff) {
if (flags & MUTEX_FLAG_HANDOFF)
break;
flags |= MUTEX_FLAG_HANDOFF;
} else {
break;
}
} else {
MUTEX_WARN_ON(flags & (MUTEX_FLAG_HANDOFF | MUTEX_FLAG_PICKUP));
task = curr;
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
}
if (atomic_long_try_cmpxchg_acquire(&lock->owner, &owner, task | flags)) {
if (task == curr)
return NULL;
break;
}
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
}
return __owner_task(owner);
}
/*
* Trylock or set HANDOFF
*/
static inline bool __mutex_trylock_or_handoff(struct mutex *lock, bool handoff)
{
return !__mutex_trylock_common(lock, handoff);
}
/*
* Actual trylock that will work on any unlocked state.
*/
static inline bool __mutex_trylock(struct mutex *lock)
{
return !__mutex_trylock_common(lock, false);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
}
#ifndef CONFIG_DEBUG_LOCK_ALLOC
/*
* Lockdep annotations are contained to the slow paths for simplicity.
* There is nothing that would stop spreading the lockdep annotations outwards
* except more code.
*/
/*
* Optimistic trylock that only works in the uncontended case. Make sure to
* follow with a __mutex_trylock() before failing.
*/
static __always_inline bool __mutex_trylock_fast(struct mutex *lock)
{
unsigned long curr = (unsigned long)current;
locking/mutex: Optimize __mutex_trylock_fast() Use try_cmpxchg to avoid the pointless TEST instruction.. And add the (missing) atomic_long_try_cmpxchg*() wrappery. On x86_64 this gives: 0000000000000710 <mutex_lock>: 0000000000000710 <mutex_lock>: 710: 65 48 8b 14 25 00 00 mov %gs:0x0,%rdx 710: 65 48 8b 14 25 00 00 mov %gs:0x0,%rdx 717: 00 00 717: 00 00 715: R_X86_64_32S current_task 715: R_X86_64_32S current_task 719: 31 c0 xor %eax,%eax 719: 31 c0 xor %eax,%eax 71b: f0 48 0f b1 17 lock cmpxchg %rdx,(%rdi) 71b: f0 48 0f b1 17 lock cmpxchg %rdx,(%rdi) 720: 48 85 c0 test %rax,%rax 720: 75 02 jne 724 <mutex_lock+0x14> 723: 75 02 jne 727 <mutex_lock+0x17> 722: f3 c3 repz retq 725: f3 c3 repz retq 724: eb da jmp 700 <__mutex_lock_slowpath> 727: eb d7 jmp 700 <__mutex_lock_slowpath> 726: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1) 729: 0f 1f 80 00 00 00 00 nopl 0x0(%rax) 72d: 00 00 00 On ARM64 this gives: 000000000000638 <mutex_lock>: 0000000000000638 <mutex_lock>: 638: d5384101 mrs x1, sp_el0 638: d5384101 mrs x1, sp_el0 63c: d2800002 mov x2, #0x0 63c: d2800002 mov x2, #0x0 640: f9800011 prfm pstl1strm, [x0] 640: f9800011 prfm pstl1strm, [x0] 644: c85ffc03 ldaxr x3, [x0] 644: c85ffc03 ldaxr x3, [x0] 648: ca020064 eor x4, x3, x2 648: ca020064 eor x4, x3, x2 64c: b5000064 cbnz x4, 658 <mutex_lock+0x20> 64c: b5000064 cbnz x4, 658 <mutex_lock+0x20> 650: c8047c01 stxr w4, x1, [x0] 650: c8047c01 stxr w4, x1, [x0] 654: 35ffff84 cbnz w4, 644 <mutex_lock+0xc> 654: 35ffff84 cbnz w4, 644 <mutex_lock+0xc> 658: b40000c3 cbz x3, 670 <mutex_lock+0x38> 658: b5000043 cbnz x3, 660 <mutex_lock+0x28> 65c: a9bf7bfd stp x29, x30, [sp,#-16]! 65c: d65f03c0 ret 660: 910003fd mov x29, sp 660: a9bf7bfd stp x29, x30, [sp,#-16]! 664: 97ffffef bl 620 <__mutex_lock_slowpath> 664: 910003fd mov x29, sp 668: a8c17bfd ldp x29, x30, [sp],#16 668: 97ffffee bl 620 <__mutex_lock_slowpath> 66c: d65f03c0 ret 66c: a8c17bfd ldp x29, x30, [sp],#16 670: d65f03c0 ret 670: d65f03c0 ret Reported-by: Matthew Wilcox <mawilcox@microsoft.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
2018-04-05 17:05:35 +08:00
unsigned long zero = 0UL;
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
locking/mutex: Optimize __mutex_trylock_fast() Use try_cmpxchg to avoid the pointless TEST instruction.. And add the (missing) atomic_long_try_cmpxchg*() wrappery. On x86_64 this gives: 0000000000000710 <mutex_lock>: 0000000000000710 <mutex_lock>: 710: 65 48 8b 14 25 00 00 mov %gs:0x0,%rdx 710: 65 48 8b 14 25 00 00 mov %gs:0x0,%rdx 717: 00 00 717: 00 00 715: R_X86_64_32S current_task 715: R_X86_64_32S current_task 719: 31 c0 xor %eax,%eax 719: 31 c0 xor %eax,%eax 71b: f0 48 0f b1 17 lock cmpxchg %rdx,(%rdi) 71b: f0 48 0f b1 17 lock cmpxchg %rdx,(%rdi) 720: 48 85 c0 test %rax,%rax 720: 75 02 jne 724 <mutex_lock+0x14> 723: 75 02 jne 727 <mutex_lock+0x17> 722: f3 c3 repz retq 725: f3 c3 repz retq 724: eb da jmp 700 <__mutex_lock_slowpath> 727: eb d7 jmp 700 <__mutex_lock_slowpath> 726: 66 2e 0f 1f 84 00 00 nopw %cs:0x0(%rax,%rax,1) 729: 0f 1f 80 00 00 00 00 nopl 0x0(%rax) 72d: 00 00 00 On ARM64 this gives: 000000000000638 <mutex_lock>: 0000000000000638 <mutex_lock>: 638: d5384101 mrs x1, sp_el0 638: d5384101 mrs x1, sp_el0 63c: d2800002 mov x2, #0x0 63c: d2800002 mov x2, #0x0 640: f9800011 prfm pstl1strm, [x0] 640: f9800011 prfm pstl1strm, [x0] 644: c85ffc03 ldaxr x3, [x0] 644: c85ffc03 ldaxr x3, [x0] 648: ca020064 eor x4, x3, x2 648: ca020064 eor x4, x3, x2 64c: b5000064 cbnz x4, 658 <mutex_lock+0x20> 64c: b5000064 cbnz x4, 658 <mutex_lock+0x20> 650: c8047c01 stxr w4, x1, [x0] 650: c8047c01 stxr w4, x1, [x0] 654: 35ffff84 cbnz w4, 644 <mutex_lock+0xc> 654: 35ffff84 cbnz w4, 644 <mutex_lock+0xc> 658: b40000c3 cbz x3, 670 <mutex_lock+0x38> 658: b5000043 cbnz x3, 660 <mutex_lock+0x28> 65c: a9bf7bfd stp x29, x30, [sp,#-16]! 65c: d65f03c0 ret 660: 910003fd mov x29, sp 660: a9bf7bfd stp x29, x30, [sp,#-16]! 664: 97ffffef bl 620 <__mutex_lock_slowpath> 664: 910003fd mov x29, sp 668: a8c17bfd ldp x29, x30, [sp],#16 668: 97ffffee bl 620 <__mutex_lock_slowpath> 66c: d65f03c0 ret 66c: a8c17bfd ldp x29, x30, [sp],#16 670: d65f03c0 ret 670: d65f03c0 ret Reported-by: Matthew Wilcox <mawilcox@microsoft.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
2018-04-05 17:05:35 +08:00
if (atomic_long_try_cmpxchg_acquire(&lock->owner, &zero, curr))
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
return true;
return false;
}
static __always_inline bool __mutex_unlock_fast(struct mutex *lock)
{
unsigned long curr = (unsigned long)current;
return atomic_long_try_cmpxchg_release(&lock->owner, &curr, 0UL);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
}
#endif
static inline void __mutex_set_flag(struct mutex *lock, unsigned long flag)
{
atomic_long_or(flag, &lock->owner);
}
static inline void __mutex_clear_flag(struct mutex *lock, unsigned long flag)
{
atomic_long_andnot(flag, &lock->owner);
}
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
static inline bool __mutex_waiter_is_first(struct mutex *lock, struct mutex_waiter *waiter)
{
return list_first_entry(&lock->wait_list, struct mutex_waiter, list) == waiter;
}
locking: Implement an algorithm choice for Wound-Wait mutexes The current Wound-Wait mutex algorithm is actually not Wound-Wait but Wait-Die. Implement also Wound-Wait as a per-ww-class choice. Wound-Wait is, contrary to Wait-Die a preemptive algorithm and is known to generate fewer backoffs. Testing reveals that this is true if the number of simultaneous contending transactions is small. As the number of simultaneous contending threads increases, Wait-Wound becomes inferior to Wait-Die in terms of elapsed time. Possibly due to the larger number of held locks of sleeping transactions. Update documentation and callers. Timings using git://people.freedesktop.org/~thomash/ww_mutex_test tag patch-18-06-15 Each thread runs 100000 batches of lock / unlock 800 ww mutexes randomly chosen out of 100000. Four core Intel x86_64: Algorithm #threads Rollbacks time Wound-Wait 4 ~100 ~17s. Wait-Die 4 ~150000 ~19s. Wound-Wait 16 ~360000 ~109s. Wait-Die 16 ~450000 ~82s. Cc: Ingo Molnar <mingo@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Gustavo Padovan <gustavo@padovan.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Sean Paul <seanpaul@chromium.org> Cc: David Airlie <airlied@linux.ie> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: linux-doc@vger.kernel.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Co-authored-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Ingo Molnar <mingo@kernel.org>
2018-06-15 16:17:38 +08:00
/*
* Add @waiter to a given location in the lock wait_list and set the
* FLAG_WAITERS flag if it's the first waiter.
*/
static void
locking: Implement an algorithm choice for Wound-Wait mutexes The current Wound-Wait mutex algorithm is actually not Wound-Wait but Wait-Die. Implement also Wound-Wait as a per-ww-class choice. Wound-Wait is, contrary to Wait-Die a preemptive algorithm and is known to generate fewer backoffs. Testing reveals that this is true if the number of simultaneous contending transactions is small. As the number of simultaneous contending threads increases, Wait-Wound becomes inferior to Wait-Die in terms of elapsed time. Possibly due to the larger number of held locks of sleeping transactions. Update documentation and callers. Timings using git://people.freedesktop.org/~thomash/ww_mutex_test tag patch-18-06-15 Each thread runs 100000 batches of lock / unlock 800 ww mutexes randomly chosen out of 100000. Four core Intel x86_64: Algorithm #threads Rollbacks time Wound-Wait 4 ~100 ~17s. Wait-Die 4 ~150000 ~19s. Wound-Wait 16 ~360000 ~109s. Wait-Die 16 ~450000 ~82s. Cc: Ingo Molnar <mingo@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Gustavo Padovan <gustavo@padovan.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Sean Paul <seanpaul@chromium.org> Cc: David Airlie <airlied@linux.ie> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: linux-doc@vger.kernel.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Co-authored-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Ingo Molnar <mingo@kernel.org>
2018-06-15 16:17:38 +08:00
__mutex_add_waiter(struct mutex *lock, struct mutex_waiter *waiter,
struct list_head *list)
{
debug_mutex_add_waiter(lock, waiter, current);
list_add_tail(&waiter->list, list);
if (__mutex_waiter_is_first(lock, waiter))
__mutex_set_flag(lock, MUTEX_FLAG_WAITERS);
}
static void
__mutex_remove_waiter(struct mutex *lock, struct mutex_waiter *waiter)
{
list_del(&waiter->list);
if (likely(list_empty(&lock->wait_list)))
__mutex_clear_flag(lock, MUTEX_FLAGS);
debug_mutex_remove_waiter(lock, waiter, current);
}
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
/*
* Give up ownership to a specific task, when @task = NULL, this is equivalent
* to a regular unlock. Sets PICKUP on a handoff, clears HANDOFF, preserves
* WAITERS. Provides RELEASE semantics like a regular unlock, the
* __mutex_trylock() provides a matching ACQUIRE semantics for the handoff.
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
*/
static void __mutex_handoff(struct mutex *lock, struct task_struct *task)
{
unsigned long owner = atomic_long_read(&lock->owner);
for (;;) {
unsigned long new;
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
MUTEX_WARN_ON(__owner_task(owner) != current);
MUTEX_WARN_ON(owner & MUTEX_FLAG_PICKUP);
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
new = (owner & MUTEX_FLAG_WAITERS);
new |= (unsigned long)task;
if (task)
new |= MUTEX_FLAG_PICKUP;
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
if (atomic_long_try_cmpxchg_release(&lock->owner, &owner, new))
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
break;
}
}
#ifndef CONFIG_DEBUG_LOCK_ALLOC
/*
* We split the mutex lock/unlock logic into separate fastpath and
* slowpath functions, to reduce the register pressure on the fastpath.
* We also put the fastpath first in the kernel image, to make sure the
* branch is predicted by the CPU as default-untaken.
*/
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
static void __sched __mutex_lock_slowpath(struct mutex *lock);
/**
* mutex_lock - acquire the mutex
* @lock: the mutex to be acquired
*
* Lock the mutex exclusively for this task. If the mutex is not
* available right now, it will sleep until it can get it.
*
* The mutex must later on be released by the same task that
* acquired it. Recursive locking is not allowed. The task
* may not exit without first unlocking the mutex. Also, kernel
* memory where the mutex resides must not be freed with
* the mutex still locked. The mutex must first be initialized
* (or statically defined) before it can be locked. memset()-ing
* the mutex to 0 is not allowed.
*
* (The CONFIG_DEBUG_MUTEXES .config option turns on debugging
* checks that will enforce the restrictions and will also do
* deadlock debugging)
*
* This function is similar to (but not equivalent to) down().
*/
void __sched mutex_lock(struct mutex *lock)
{
might_sleep();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
if (!__mutex_trylock_fast(lock))
__mutex_lock_slowpath(lock);
}
EXPORT_SYMBOL(mutex_lock);
#endif
#include "ww_mutex.h"
#ifdef CONFIG_MUTEX_SPIN_ON_OWNER
/*
* Trylock variant that returns the owning task on failure.
*/
static inline struct task_struct *__mutex_trylock_or_owner(struct mutex *lock)
{
return __mutex_trylock_common(lock, false);
}
static inline
bool ww_mutex_spin_on_owner(struct mutex *lock, struct ww_acquire_ctx *ww_ctx,
struct mutex_waiter *waiter)
{
struct ww_mutex *ww;
ww = container_of(lock, struct ww_mutex, base);
/*
* If ww->ctx is set the contents are undefined, only
* by acquiring wait_lock there is a guarantee that
* they are not invalid when reading.
*
* As such, when deadlock detection needs to be
* performed the optimistic spinning cannot be done.
*
* Check this in every inner iteration because we may
* be racing against another thread's ww_mutex_lock.
*/
if (ww_ctx->acquired > 0 && READ_ONCE(ww->ctx))
return false;
/*
* If we aren't on the wait list yet, cancel the spin
* if there are waiters. We want to avoid stealing the
* lock from a waiter with an earlier stamp, since the
* other thread may already own a lock that we also
* need.
*/
if (!waiter && (atomic_long_read(&lock->owner) & MUTEX_FLAG_WAITERS))
return false;
/*
* Similarly, stop spinning if we are no longer the
* first waiter.
*/
if (waiter && !__mutex_waiter_is_first(lock, waiter))
return false;
return true;
}
/*
locking/ww_mutex: Re-check ww->ctx in the inner optimistic spin loop In the following scenario, thread #1 should back off its attempt to lock ww1 and unlock ww2 (assuming the acquire context stamps are ordered accordingly). Thread #0 Thread #1 --------- --------- successfully lock ww2 set ww1->base.owner attempt to lock ww1 confirm ww1->ctx == NULL enter mutex_spin_on_owner set ww1->ctx What was likely to happen previously is: attempt to lock ww2 refuse to spin because ww2->ctx != NULL schedule() detect thread #0 is off CPU stop optimistic spin return -EDEADLK unlock ww2 wakeup thread #0 lock ww2 Now, we are more likely to see: detect ww1->ctx != NULL stop optimistic spin return -EDEADLK unlock ww2 successfully lock ww2 ... because thread #1 will stop its optimistic spin as soon as possible. The whole scenario is quite unlikely, since it requires thread #1 to get between thread #0 setting the owner and setting the ctx. But since we're idling here anyway, the additional check is basically free. Found by inspection. Signed-off-by: Nicolai Hähnle <Nicolai.Haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-10-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:37 +08:00
* Look out! "owner" is an entirely speculative pointer access and not
* reliable.
*
* "noinline" so that this function shows up on perf profiles.
*/
static noinline
locking/ww_mutex: Re-check ww->ctx in the inner optimistic spin loop In the following scenario, thread #1 should back off its attempt to lock ww1 and unlock ww2 (assuming the acquire context stamps are ordered accordingly). Thread #0 Thread #1 --------- --------- successfully lock ww2 set ww1->base.owner attempt to lock ww1 confirm ww1->ctx == NULL enter mutex_spin_on_owner set ww1->ctx What was likely to happen previously is: attempt to lock ww2 refuse to spin because ww2->ctx != NULL schedule() detect thread #0 is off CPU stop optimistic spin return -EDEADLK unlock ww2 wakeup thread #0 lock ww2 Now, we are more likely to see: detect ww1->ctx != NULL stop optimistic spin return -EDEADLK unlock ww2 successfully lock ww2 ... because thread #1 will stop its optimistic spin as soon as possible. The whole scenario is quite unlikely, since it requires thread #1 to get between thread #0 setting the owner and setting the ctx. But since we're idling here anyway, the additional check is basically free. Found by inspection. Signed-off-by: Nicolai Hähnle <Nicolai.Haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-10-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:37 +08:00
bool mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner,
struct ww_acquire_ctx *ww_ctx, struct mutex_waiter *waiter)
{
bool ret = true;
rcu_read_lock();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
while (__mutex_owner(lock) == owner) {
/*
* Ensure we emit the owner->on_cpu, dereference _after_
* checking lock->owner still matches owner. If that fails,
* owner might point to freed memory. If it still matches,
* the rcu_read_lock() ensures the memory stays valid.
*/
barrier();
locking/mutex: Break out of expensive busy-loop on {mutex,rwsem}_spin_on_owner() when owner vCPU is preempted An over-committed guest with more vCPUs than pCPUs has a heavy overload in the two spin_on_owner. This blames on the lock holder preemption issue. Break out of the loop if the vCPU is preempted: if vcpu_is_preempted(cpu) is true. test-case: perf record -a perf bench sched messaging -g 400 -p && perf report before patch: 20.68% sched-messaging [kernel.vmlinux] [k] mutex_spin_on_owner 8.45% sched-messaging [kernel.vmlinux] [k] mutex_unlock 4.12% sched-messaging [kernel.vmlinux] [k] system_call 3.01% sched-messaging [kernel.vmlinux] [k] system_call_common 2.83% sched-messaging [kernel.vmlinux] [k] copypage_power7 2.64% sched-messaging [kernel.vmlinux] [k] rwsem_spin_on_owner 2.00% sched-messaging [kernel.vmlinux] [k] osq_lock after patch: 9.99% sched-messaging [kernel.vmlinux] [k] mutex_unlock 5.28% sched-messaging [unknown] [H] 0xc0000000000768e0 4.27% sched-messaging [kernel.vmlinux] [k] __copy_tofrom_user_power7 3.77% sched-messaging [kernel.vmlinux] [k] copypage_power7 3.24% sched-messaging [kernel.vmlinux] [k] _raw_write_lock_irq 3.02% sched-messaging [kernel.vmlinux] [k] system_call 2.69% sched-messaging [kernel.vmlinux] [k] wait_consider_task Tested-by: Juergen Gross <jgross@suse.com> Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Cc: David.Laight@ACULAB.COM Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: benh@kernel.crashing.org Cc: boqun.feng@gmail.com Cc: bsingharora@gmail.com Cc: dave@stgolabs.net Cc: kernellwp@gmail.com Cc: konrad.wilk@oracle.com Cc: linuxppc-dev@lists.ozlabs.org Cc: mpe@ellerman.id.au Cc: paulmck@linux.vnet.ibm.com Cc: paulus@samba.org Cc: rkrcmar@redhat.com Cc: virtualization@lists.linux-foundation.org Cc: will.deacon@arm.com Cc: xen-devel-request@lists.xenproject.org Cc: xen-devel@lists.xenproject.org Link: http://lkml.kernel.org/r/1478077718-37424-4-git-send-email-xinhui.pan@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-02 17:08:30 +08:00
/*
* Use vcpu_is_preempted to detect lock holder preemption issue.
*/
if (!owner->on_cpu || need_resched() ||
vcpu_is_preempted(task_cpu(owner))) {
ret = false;
break;
}
if (ww_ctx && !ww_mutex_spin_on_owner(lock, ww_ctx, waiter)) {
ret = false;
break;
locking/ww_mutex: Re-check ww->ctx in the inner optimistic spin loop In the following scenario, thread #1 should back off its attempt to lock ww1 and unlock ww2 (assuming the acquire context stamps are ordered accordingly). Thread #0 Thread #1 --------- --------- successfully lock ww2 set ww1->base.owner attempt to lock ww1 confirm ww1->ctx == NULL enter mutex_spin_on_owner set ww1->ctx What was likely to happen previously is: attempt to lock ww2 refuse to spin because ww2->ctx != NULL schedule() detect thread #0 is off CPU stop optimistic spin return -EDEADLK unlock ww2 wakeup thread #0 lock ww2 Now, we are more likely to see: detect ww1->ctx != NULL stop optimistic spin return -EDEADLK unlock ww2 successfully lock ww2 ... because thread #1 will stop its optimistic spin as soon as possible. The whole scenario is quite unlikely, since it requires thread #1 to get between thread #0 setting the owner and setting the ctx. But since we're idling here anyway, the additional check is basically free. Found by inspection. Signed-off-by: Nicolai Hähnle <Nicolai.Haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-10-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:37 +08:00
}
cpu_relax();
}
rcu_read_unlock();
return ret;
}
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
/*
* Initial check for entering the mutex spinning loop
*/
static inline int mutex_can_spin_on_owner(struct mutex *lock)
{
struct task_struct *owner;
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
int retval = 1;
if (need_resched())
return 0;
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
rcu_read_lock();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
owner = __mutex_owner(lock);
locking/mutex: Break out of expensive busy-loop on {mutex,rwsem}_spin_on_owner() when owner vCPU is preempted An over-committed guest with more vCPUs than pCPUs has a heavy overload in the two spin_on_owner. This blames on the lock holder preemption issue. Break out of the loop if the vCPU is preempted: if vcpu_is_preempted(cpu) is true. test-case: perf record -a perf bench sched messaging -g 400 -p && perf report before patch: 20.68% sched-messaging [kernel.vmlinux] [k] mutex_spin_on_owner 8.45% sched-messaging [kernel.vmlinux] [k] mutex_unlock 4.12% sched-messaging [kernel.vmlinux] [k] system_call 3.01% sched-messaging [kernel.vmlinux] [k] system_call_common 2.83% sched-messaging [kernel.vmlinux] [k] copypage_power7 2.64% sched-messaging [kernel.vmlinux] [k] rwsem_spin_on_owner 2.00% sched-messaging [kernel.vmlinux] [k] osq_lock after patch: 9.99% sched-messaging [kernel.vmlinux] [k] mutex_unlock 5.28% sched-messaging [unknown] [H] 0xc0000000000768e0 4.27% sched-messaging [kernel.vmlinux] [k] __copy_tofrom_user_power7 3.77% sched-messaging [kernel.vmlinux] [k] copypage_power7 3.24% sched-messaging [kernel.vmlinux] [k] _raw_write_lock_irq 3.02% sched-messaging [kernel.vmlinux] [k] system_call 2.69% sched-messaging [kernel.vmlinux] [k] wait_consider_task Tested-by: Juergen Gross <jgross@suse.com> Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Cc: David.Laight@ACULAB.COM Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: benh@kernel.crashing.org Cc: boqun.feng@gmail.com Cc: bsingharora@gmail.com Cc: dave@stgolabs.net Cc: kernellwp@gmail.com Cc: konrad.wilk@oracle.com Cc: linuxppc-dev@lists.ozlabs.org Cc: mpe@ellerman.id.au Cc: paulmck@linux.vnet.ibm.com Cc: paulus@samba.org Cc: rkrcmar@redhat.com Cc: virtualization@lists.linux-foundation.org Cc: will.deacon@arm.com Cc: xen-devel-request@lists.xenproject.org Cc: xen-devel@lists.xenproject.org Link: http://lkml.kernel.org/r/1478077718-37424-4-git-send-email-xinhui.pan@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-02 17:08:30 +08:00
/*
* As lock holder preemption issue, we both skip spinning if task is not
* on cpu or its cpu is preempted
*/
if (owner)
locking/mutex: Break out of expensive busy-loop on {mutex,rwsem}_spin_on_owner() when owner vCPU is preempted An over-committed guest with more vCPUs than pCPUs has a heavy overload in the two spin_on_owner. This blames on the lock holder preemption issue. Break out of the loop if the vCPU is preempted: if vcpu_is_preempted(cpu) is true. test-case: perf record -a perf bench sched messaging -g 400 -p && perf report before patch: 20.68% sched-messaging [kernel.vmlinux] [k] mutex_spin_on_owner 8.45% sched-messaging [kernel.vmlinux] [k] mutex_unlock 4.12% sched-messaging [kernel.vmlinux] [k] system_call 3.01% sched-messaging [kernel.vmlinux] [k] system_call_common 2.83% sched-messaging [kernel.vmlinux] [k] copypage_power7 2.64% sched-messaging [kernel.vmlinux] [k] rwsem_spin_on_owner 2.00% sched-messaging [kernel.vmlinux] [k] osq_lock after patch: 9.99% sched-messaging [kernel.vmlinux] [k] mutex_unlock 5.28% sched-messaging [unknown] [H] 0xc0000000000768e0 4.27% sched-messaging [kernel.vmlinux] [k] __copy_tofrom_user_power7 3.77% sched-messaging [kernel.vmlinux] [k] copypage_power7 3.24% sched-messaging [kernel.vmlinux] [k] _raw_write_lock_irq 3.02% sched-messaging [kernel.vmlinux] [k] system_call 2.69% sched-messaging [kernel.vmlinux] [k] wait_consider_task Tested-by: Juergen Gross <jgross@suse.com> Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Acked-by: Paolo Bonzini <pbonzini@redhat.com> Cc: David.Laight@ACULAB.COM Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: benh@kernel.crashing.org Cc: boqun.feng@gmail.com Cc: bsingharora@gmail.com Cc: dave@stgolabs.net Cc: kernellwp@gmail.com Cc: konrad.wilk@oracle.com Cc: linuxppc-dev@lists.ozlabs.org Cc: mpe@ellerman.id.au Cc: paulmck@linux.vnet.ibm.com Cc: paulus@samba.org Cc: rkrcmar@redhat.com Cc: virtualization@lists.linux-foundation.org Cc: will.deacon@arm.com Cc: xen-devel-request@lists.xenproject.org Cc: xen-devel@lists.xenproject.org Link: http://lkml.kernel.org/r/1478077718-37424-4-git-send-email-xinhui.pan@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-11-02 17:08:30 +08:00
retval = owner->on_cpu && !vcpu_is_preempted(task_cpu(owner));
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
rcu_read_unlock();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
/*
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
* If lock->owner is not set, the mutex has been released. Return true
* such that we'll trylock in the spin path, which is a faster option
* than the blocking slow path.
mutex: Queue mutex spinners with MCS lock to reduce cacheline contention The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option turned on) allow multiple tasks to spin on a single mutex concurrently. A potential problem with the current approach is that when the mutex becomes available, all the spinning tasks will try to acquire the mutex more or less simultaneously. As a result, there will be a lot of cacheline bouncing especially on systems with a large number of CPUs. This patch tries to reduce this kind of contention by putting the mutex spinners into a queue so that only the first one in the queue will try to acquire the mutex. This will reduce contention and allow all the tasks to move forward faster. The queuing of mutex spinners is done using an MCS lock based implementation which will further reduce contention on the mutex cacheline than a similar ticket spinlock based implementation. This patch will add a new field into the mutex data structure for holding the MCS lock. This expands the mutex size by 8 bytes for 64-bit system and 4 bytes for 32-bit system. This overhead will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off. The following table shows the jobs per minute (JPM) scalability data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The numactl command is used to restrict the running of the fserver workloads to 1/2/4/8 nodes with hyperthreading off. +-----------------+-----------+-----------+-------------+----------+ | Configuration | Mean JPM | Mean JPM | Mean JPM | % Change | | | w/o patch | patch 1 | patches 1&2 | 1->1&2 | +-----------------+------------------------------------------------+ | | User Range 1100 - 2000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% | | 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% | | 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% | | 1 node , HT off | 198141 | 197972 | 198075 | +0.1% | +-----------------+------------------------------------------------+ | | User Range 200 - 1000 | +-----------------+------------------------------------------------+ | 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% | | 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% | | 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% | | 1 node , HT off | 197588 | 197622 | 197582 | 0.0% | +-----------------+-----------+-----------+-------------+----------+ At low user range 10-100, the JPM differences were within +/-1%. So they are not that interesting. The fserver workload uses mutex spinning extensively. With just the mutex change in the first patch, there is no noticeable change in performance. Rather, there is a slight drop in performance. This mutex spinning patch more than recovers the lost performance and show a significant increase of +30% at high user load with the full 8 nodes. Similar improvements were also seen in a 3.8 kernel. The table below shows the %time spent by different kernel functions as reported by perf when running the fserver workload at 1500 users with all 8 nodes. +-----------------------+-----------+---------+-------------+ | Function | % time | % time | % time | | | w/o patch | patch 1 | patches 1&2 | +-----------------------+-----------+---------+-------------+ | __read_lock_failed | 34.96% | 34.91% | 29.14% | | __write_lock_failed | 10.14% | 10.68% | 7.51% | | mutex_spin_on_owner | 3.62% | 3.42% | 2.33% | | mspin_lock | N/A | N/A | 9.90% | | __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% | | _raw_spin_lock | 2.25% | 2.50% | 1.10% | +-----------------------+-----------+---------+-------------+ The fserver workload for an 8-node system is dominated by the contention in the read/write lock. Mutex contention also plays a role. With the first patch only, mutex contention is down (as shown by the __mutex_lock_slowpath figure) which help a little bit. We saw only a few percents improvement with that. By applying patch 2 as well, the single mutex_spin_on_owner figure is now split out into an additional mspin_lock figure. The time increases from 3.42% to 11.23%. It shows a great reduction in contention among the spinners leading to a 30% improvement. The time ratio 9.9/2.33=4.3 indicates that there are on average 4+ spinners waiting in the spin_lock loop for each spinner in the mutex_spin_on_owner loop. Contention in other locking functions also go down by quite a lot. The table below shows the performance change of both patches 1 & 2 over patch 1 alone in other AIM7 workloads (at 8 nodes, hyperthreading off). +--------------+---------------+----------------+-----------------+ | Workload | mean % change | mean % change | mean % change | | | 10-100 users | 200-1000 users | 1100-2000 users | +--------------+---------------+----------------+-----------------+ | alltests | 0.0% | -0.8% | +0.6% | | five_sec | -0.3% | +0.8% | +0.8% | | high_systime | +0.4% | +2.4% | +2.1% | | new_fserver | +0.1% | +14.1% | +34.2% | | shared | -0.5% | -0.3% | -0.4% | | short | -1.7% | -9.8% | -8.3% | +--------------+---------------+----------------+-----------------+ The short workload is the only one that shows a decline in performance probably due to the spinner locking and queuing overhead. Signed-off-by: Waiman Long <Waiman.Long@hp.com> Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Chandramouleeswaran Aswin <aswin@hp.com> Cc: Norton Scott J <scott.norton@hp.com> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: David Howells <dhowells@redhat.com> Cc: Dave Jones <davej@redhat.com> Cc: Clark Williams <williams@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-04-18 03:23:13 +08:00
*/
return retval;
}
/*
* Optimistic spinning.
*
* We try to spin for acquisition when we find that the lock owner
* is currently running on a (different) CPU and while we don't
* need to reschedule. The rationale is that if the lock owner is
* running, it is likely to release the lock soon.
*
* The mutex spinners are queued up using MCS lock so that only one
* spinner can compete for the mutex. However, if mutex spinning isn't
* going to happen, there is no point in going through the lock/unlock
* overhead.
*
* Returns true when the lock was taken, otherwise false, indicating
* that we need to jump to the slowpath and sleep.
*
* The waiter flag is set to true if the spinner is a waiter in the wait
* queue. The waiter-spinner will spin on the lock directly and concurrently
* with the spinner at the head of the OSQ, if present, until the owner is
* changed to itself.
*/
static __always_inline bool
mutex_optimistic_spin(struct mutex *lock, struct ww_acquire_ctx *ww_ctx,
struct mutex_waiter *waiter)
{
if (!waiter) {
/*
* The purpose of the mutex_can_spin_on_owner() function is
* to eliminate the overhead of osq_lock() and osq_unlock()
* in case spinning isn't possible. As a waiter-spinner
* is not going to take OSQ lock anyway, there is no need
* to call mutex_can_spin_on_owner().
*/
if (!mutex_can_spin_on_owner(lock))
goto fail;
/*
* In order to avoid a stampede of mutex spinners trying to
* acquire the mutex all at once, the spinners need to take a
* MCS (queued) lock first before spinning on the owner field.
*/
if (!osq_lock(&lock->osq))
goto fail;
}
for (;;) {
struct task_struct *owner;
/* Try to acquire the mutex... */
owner = __mutex_trylock_or_owner(lock);
if (!owner)
break;
/*
* There's an owner, wait for it to either
* release the lock or go to sleep.
*/
if (!mutex_spin_on_owner(lock, owner, ww_ctx, waiter))
goto fail_unlock;
/*
* The cpu_relax() call is a compiler barrier which forces
* everything in this loop to be re-loaded. We don't need
* memory barriers as we'll eventually observe the right
* values at the cost of a few extra spins.
*/
cpu_relax();
}
if (!waiter)
osq_unlock(&lock->osq);
return true;
fail_unlock:
if (!waiter)
osq_unlock(&lock->osq);
fail:
/*
* If we fell out of the spin path because of need_resched(),
* reschedule now, before we try-lock the mutex. This avoids getting
* scheduled out right after we obtained the mutex.
*/
if (need_resched()) {
/*
* We _should_ have TASK_RUNNING here, but just in case
* we do not, make it so, otherwise we might get stuck.
*/
__set_current_state(TASK_RUNNING);
schedule_preempt_disabled();
}
return false;
}
#else
static __always_inline bool
mutex_optimistic_spin(struct mutex *lock, struct ww_acquire_ctx *ww_ctx,
struct mutex_waiter *waiter)
{
return false;
}
#endif
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigned long ip);
/**
* mutex_unlock - release the mutex
* @lock: the mutex to be released
*
* Unlock a mutex that has been locked by this task previously.
*
* This function must not be used in interrupt context. Unlocking
* of a not locked mutex is not allowed.
*
* This function is similar to (but not equivalent to) up().
*/
void __sched mutex_unlock(struct mutex *lock)
{
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
#ifndef CONFIG_DEBUG_LOCK_ALLOC
if (__mutex_unlock_fast(lock))
return;
mutex: implement adaptive spinning Change mutex contention behaviour such that it will sometimes busy wait on acquisition - moving its behaviour closer to that of spinlocks. This concept got ported to mainline from the -rt tree, where it was originally implemented for rtmutexes by Steven Rostedt, based on work by Gregory Haskins. Testing with Ingo's test-mutex application (http://lkml.org/lkml/2006/1/8/50) gave a 345% boost for VFS scalability on my testbox: # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 296604 # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 85870 The key criteria for the busy wait is that the lock owner has to be running on a (different) cpu. The idea is that as long as the owner is running, there is a fair chance it'll release the lock soon, and thus we'll be better off spinning instead of blocking/scheduling. Since regular mutexes (as opposed to rtmutexes) do not atomically track the owner, we add the owner in a non-atomic fashion and deal with the races in the slowpath. Furthermore, to ease the testing of the performance impact of this new code, there is means to disable this behaviour runtime (without having to reboot the system), when scheduler debugging is enabled (CONFIG_SCHED_DEBUG=y), by issuing the following command: # echo NO_OWNER_SPIN > /debug/sched_features This command re-enables spinning again (this is also the default): # echo OWNER_SPIN > /debug/sched_features Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-01-12 21:01:47 +08:00
#endif
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
__mutex_unlock_slowpath(lock, _RET_IP_);
}
EXPORT_SYMBOL(mutex_unlock);
/**
* ww_mutex_unlock - release the w/w mutex
* @lock: the mutex to be released
*
* Unlock a mutex that has been locked by this task previously with any of the
* ww_mutex_lock* functions (with or without an acquire context). It is
* forbidden to release the locks after releasing the acquire context.
*
* This function must not be used in interrupt context. Unlocking
* of a unlocked mutex is not allowed.
*/
void __sched ww_mutex_unlock(struct ww_mutex *lock)
{
__ww_mutex_unlock(lock);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
mutex_unlock(&lock->base);
}
EXPORT_SYMBOL(ww_mutex_unlock);
/*
* Lock a mutex (possibly interruptible), slowpath:
*/
static __always_inline int __sched
__mutex_lock_common(struct mutex *lock, unsigned int state, unsigned int subclass,
struct lockdep_map *nest_lock, unsigned long ip,
struct ww_acquire_ctx *ww_ctx, const bool use_ww_ctx)
{
struct mutex_waiter waiter;
struct ww_mutex *ww;
int ret;
if (!use_ww_ctx)
ww_ctx = NULL;
might_sleep();
MUTEX_WARN_ON(lock->magic != lock);
ww = container_of(lock, struct ww_mutex, base);
if (ww_ctx) {
if (unlikely(ww_ctx == READ_ONCE(ww->ctx)))
return -EALREADY;
locking: Implement an algorithm choice for Wound-Wait mutexes The current Wound-Wait mutex algorithm is actually not Wound-Wait but Wait-Die. Implement also Wound-Wait as a per-ww-class choice. Wound-Wait is, contrary to Wait-Die a preemptive algorithm and is known to generate fewer backoffs. Testing reveals that this is true if the number of simultaneous contending transactions is small. As the number of simultaneous contending threads increases, Wait-Wound becomes inferior to Wait-Die in terms of elapsed time. Possibly due to the larger number of held locks of sleeping transactions. Update documentation and callers. Timings using git://people.freedesktop.org/~thomash/ww_mutex_test tag patch-18-06-15 Each thread runs 100000 batches of lock / unlock 800 ww mutexes randomly chosen out of 100000. Four core Intel x86_64: Algorithm #threads Rollbacks time Wound-Wait 4 ~100 ~17s. Wait-Die 4 ~150000 ~19s. Wound-Wait 16 ~360000 ~109s. Wait-Die 16 ~450000 ~82s. Cc: Ingo Molnar <mingo@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Gustavo Padovan <gustavo@padovan.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Sean Paul <seanpaul@chromium.org> Cc: David Airlie <airlied@linux.ie> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: linux-doc@vger.kernel.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Co-authored-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Ingo Molnar <mingo@kernel.org>
2018-06-15 16:17:38 +08:00
/*
* Reset the wounded flag after a kill. No other process can
* race and wound us here since they can't have a valid owner
* pointer if we don't have any locks held.
*/
if (ww_ctx->acquired == 0)
ww_ctx->wounded = 0;
#ifdef CONFIG_DEBUG_LOCK_ALLOC
nest_lock = &ww_ctx->dep_map;
#endif
}
preempt_disable();
mutex_acquire_nest(&lock->dep_map, subclass, 0, nest_lock, ip);
if (__mutex_trylock(lock) ||
mutex_optimistic_spin(lock, ww_ctx, NULL)) {
/* got the lock, yay! */
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
lock_acquired(&lock->dep_map, ip);
if (ww_ctx)
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
ww_mutex_set_context_fastpath(ww, ww_ctx);
preempt_enable();
return 0;
mutex: implement adaptive spinning Change mutex contention behaviour such that it will sometimes busy wait on acquisition - moving its behaviour closer to that of spinlocks. This concept got ported to mainline from the -rt tree, where it was originally implemented for rtmutexes by Steven Rostedt, based on work by Gregory Haskins. Testing with Ingo's test-mutex application (http://lkml.org/lkml/2006/1/8/50) gave a 345% boost for VFS scalability on my testbox: # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 296604 # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 85870 The key criteria for the busy wait is that the lock owner has to be running on a (different) cpu. The idea is that as long as the owner is running, there is a fair chance it'll release the lock soon, and thus we'll be better off spinning instead of blocking/scheduling. Since regular mutexes (as opposed to rtmutexes) do not atomically track the owner, we add the owner in a non-atomic fashion and deal with the races in the slowpath. Furthermore, to ease the testing of the performance impact of this new code, there is means to disable this behaviour runtime (without having to reboot the system), when scheduler debugging is enabled (CONFIG_SCHED_DEBUG=y), by issuing the following command: # echo NO_OWNER_SPIN > /debug/sched_features This command re-enables spinning again (this is also the default): # echo OWNER_SPIN > /debug/sched_features Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-01-12 21:01:47 +08:00
}
raw_spin_lock(&lock->wait_lock);
/*
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
* After waiting to acquire the wait_lock, try again.
*/
if (__mutex_trylock(lock)) {
if (ww_ctx)
__ww_mutex_check_waiters(lock, ww_ctx);
mutex: Do not unnecessarily deal with waiters Upon entering the slowpath, we immediately attempt to acquire the lock by checking if it is already unlocked. If we are lucky enough that this is the case, then we don't need to deal with any waiter related logic. Furthermore any checks for an empty wait_list are unnecessary as we already know that count is non-negative and hence no one is waiting for the lock. Move the count check and xchg calls to be done before any waiters are setup - including waiter debugging. Upon failure to acquire the lock, the xchg sets the counter to 0, instead of -1 as it was originally. This can be done here since we set it back to -1 right at the beginning of the loop so other waiters are woken up when the lock is released. When tested on a 8-socket (80 core) system against a vanilla 3.10-rc1 kernel, this patch provides some small performance benefits (+2-6%). While it could be considered in the noise level, the average percentages were stable across multiple runs and no performance regressions were seen. Two big winners, for small amounts of users (10-100), were the short and compute workloads had a +19.36% and +%15.76% in jobs per minute. Also change some break statements to 'goto slowpath', which IMO makes a little more intuitive to read. Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Acked-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1372450398.2106.1.camel@buesod1.americas.hpqcorp.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-29 04:13:18 +08:00
goto skip_wait;
}
mutex: Do not unnecessarily deal with waiters Upon entering the slowpath, we immediately attempt to acquire the lock by checking if it is already unlocked. If we are lucky enough that this is the case, then we don't need to deal with any waiter related logic. Furthermore any checks for an empty wait_list are unnecessary as we already know that count is non-negative and hence no one is waiting for the lock. Move the count check and xchg calls to be done before any waiters are setup - including waiter debugging. Upon failure to acquire the lock, the xchg sets the counter to 0, instead of -1 as it was originally. This can be done here since we set it back to -1 right at the beginning of the loop so other waiters are woken up when the lock is released. When tested on a 8-socket (80 core) system against a vanilla 3.10-rc1 kernel, this patch provides some small performance benefits (+2-6%). While it could be considered in the noise level, the average percentages were stable across multiple runs and no performance regressions were seen. Two big winners, for small amounts of users (10-100), were the short and compute workloads had a +19.36% and +%15.76% in jobs per minute. Also change some break statements to 'goto slowpath', which IMO makes a little more intuitive to read. Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Acked-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1372450398.2106.1.camel@buesod1.americas.hpqcorp.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-29 04:13:18 +08:00
debug_mutex_lock_common(lock, &waiter);
waiter.task = current;
if (ww_ctx)
waiter.ww_ctx = ww_ctx;
locking/ww_mutex: Add waiters in stamp order Add regular waiters in stamp order. Keep adding waiters that have no context in FIFO order and take care not to starve them. While adding our task as a waiter, back off if we detect that there is a waiter with a lower stamp in front of us. Make sure to call lock_contended even when we back off early. For w/w mutexes, being first in the wait list is only stable when taking the lock without a context. Therefore, the purpose of the first flag is split into two: 'first' remains to indicate whether we want to spin optimistically, while 'handoff' indicates that we should be prepared to accept a handoff. For w/w locking with a context, we always accept handoffs after the first schedule(), to handle the following sequence of events: 1. Task #0 unlocks and hands off to Task #2 which is first in line 2. Task #1 adds itself in front of Task #2 3. Task #2 wakes up and must accept the handoff even though it is no longer first in line Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: =?UTF-8?q?Nicolai=20H=C3=A4hnle?= <Nicolai.Haehnle@amd.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-7-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:34 +08:00
lock_contended(&lock->dep_map, ip);
if (!use_ww_ctx) {
/* add waiting tasks to the end of the waitqueue (FIFO): */
locking: Implement an algorithm choice for Wound-Wait mutexes The current Wound-Wait mutex algorithm is actually not Wound-Wait but Wait-Die. Implement also Wound-Wait as a per-ww-class choice. Wound-Wait is, contrary to Wait-Die a preemptive algorithm and is known to generate fewer backoffs. Testing reveals that this is true if the number of simultaneous contending transactions is small. As the number of simultaneous contending threads increases, Wait-Wound becomes inferior to Wait-Die in terms of elapsed time. Possibly due to the larger number of held locks of sleeping transactions. Update documentation and callers. Timings using git://people.freedesktop.org/~thomash/ww_mutex_test tag patch-18-06-15 Each thread runs 100000 batches of lock / unlock 800 ww mutexes randomly chosen out of 100000. Four core Intel x86_64: Algorithm #threads Rollbacks time Wound-Wait 4 ~100 ~17s. Wait-Die 4 ~150000 ~19s. Wound-Wait 16 ~360000 ~109s. Wait-Die 16 ~450000 ~82s. Cc: Ingo Molnar <mingo@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Gustavo Padovan <gustavo@padovan.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Sean Paul <seanpaul@chromium.org> Cc: David Airlie <airlied@linux.ie> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: linux-doc@vger.kernel.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Co-authored-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Ingo Molnar <mingo@kernel.org>
2018-06-15 16:17:38 +08:00
__mutex_add_waiter(lock, &waiter, &lock->wait_list);
locking/ww_mutex: Add waiters in stamp order Add regular waiters in stamp order. Keep adding waiters that have no context in FIFO order and take care not to starve them. While adding our task as a waiter, back off if we detect that there is a waiter with a lower stamp in front of us. Make sure to call lock_contended even when we back off early. For w/w mutexes, being first in the wait list is only stable when taking the lock without a context. Therefore, the purpose of the first flag is split into two: 'first' remains to indicate whether we want to spin optimistically, while 'handoff' indicates that we should be prepared to accept a handoff. For w/w locking with a context, we always accept handoffs after the first schedule(), to handle the following sequence of events: 1. Task #0 unlocks and hands off to Task #2 which is first in line 2. Task #1 adds itself in front of Task #2 3. Task #2 wakes up and must accept the handoff even though it is no longer first in line Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: =?UTF-8?q?Nicolai=20H=C3=A4hnle?= <Nicolai.Haehnle@amd.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-7-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:34 +08:00
} else {
/*
* Add in stamp order, waking up waiters that must kill
* themselves.
*/
locking/ww_mutex: Add waiters in stamp order Add regular waiters in stamp order. Keep adding waiters that have no context in FIFO order and take care not to starve them. While adding our task as a waiter, back off if we detect that there is a waiter with a lower stamp in front of us. Make sure to call lock_contended even when we back off early. For w/w mutexes, being first in the wait list is only stable when taking the lock without a context. Therefore, the purpose of the first flag is split into two: 'first' remains to indicate whether we want to spin optimistically, while 'handoff' indicates that we should be prepared to accept a handoff. For w/w locking with a context, we always accept handoffs after the first schedule(), to handle the following sequence of events: 1. Task #0 unlocks and hands off to Task #2 which is first in line 2. Task #1 adds itself in front of Task #2 3. Task #2 wakes up and must accept the handoff even though it is no longer first in line Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: =?UTF-8?q?Nicolai=20H=C3=A4hnle?= <Nicolai.Haehnle@amd.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-7-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:34 +08:00
ret = __ww_mutex_add_waiter(&waiter, lock, ww_ctx);
if (ret)
goto err_early_kill;
locking/ww_mutex: Add waiters in stamp order Add regular waiters in stamp order. Keep adding waiters that have no context in FIFO order and take care not to starve them. While adding our task as a waiter, back off if we detect that there is a waiter with a lower stamp in front of us. Make sure to call lock_contended even when we back off early. For w/w mutexes, being first in the wait list is only stable when taking the lock without a context. Therefore, the purpose of the first flag is split into two: 'first' remains to indicate whether we want to spin optimistically, while 'handoff' indicates that we should be prepared to accept a handoff. For w/w locking with a context, we always accept handoffs after the first schedule(), to handle the following sequence of events: 1. Task #0 unlocks and hands off to Task #2 which is first in line 2. Task #1 adds itself in front of Task #2 3. Task #2 wakes up and must accept the handoff even though it is no longer first in line Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: =?UTF-8?q?Nicolai=20H=C3=A4hnle?= <Nicolai.Haehnle@amd.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Maarten Lankhorst <dev@mblankhorst.nl> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dri-devel@lists.freedesktop.org Link: http://lkml.kernel.org/r/1482346000-9927-7-git-send-email-nhaehnle@gmail.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-12-22 02:46:34 +08:00
}
sched/core: Remove set_task_state() This is a nasty interface and setting the state of a foreign task must not be done. As of the following commit: be628be0956 ("bcache: Make gc wakeup sane, remove set_task_state()") ... everyone in the kernel calls set_task_state() with current, allowing the helper to be removed. However, as the comment indicates, it is still around for those archs where computing current is more expensive than using a pointer, at least in theory. An important arch that is affected is arm64, however this has been addressed now [1] and performance is up to par making no difference with either calls. Of all the callers, if any, it's the locking bits that would care most about this -- ie: we end up passing a tsk pointer to a lot of the lock slowpath, and setting ->state on that. The following numbers are based on two tests: a custom ad-hoc microbenchmark that just measures latencies (for ~65 million calls) between get_task_state() vs get_current_state(). Secondly for a higher overview, an unlink microbenchmark was used, which pounds on a single file with open, close,unlink combos with increasing thread counts (up to 4x ncpus). While the workload is quite unrealistic, it does contend a lot on the inode mutex or now rwsem. [1] https://lkml.kernel.org/r/1483468021-8237-1-git-send-email-mark.rutland@arm.com == 1. x86-64 == Avg runtime set_task_state(): 601 msecs Avg runtime set_current_state(): 552 msecs vanilla dirty Hmean unlink1-processes-2 36089.26 ( 0.00%) 38977.33 ( 8.00%) Hmean unlink1-processes-5 28555.01 ( 0.00%) 29832.55 ( 4.28%) Hmean unlink1-processes-8 37323.75 ( 0.00%) 44974.57 ( 20.50%) Hmean unlink1-processes-12 43571.88 ( 0.00%) 44283.01 ( 1.63%) Hmean unlink1-processes-21 34431.52 ( 0.00%) 38284.45 ( 11.19%) Hmean unlink1-processes-30 34813.26 ( 0.00%) 37975.17 ( 9.08%) Hmean unlink1-processes-48 37048.90 ( 0.00%) 39862.78 ( 7.59%) Hmean unlink1-processes-79 35630.01 ( 0.00%) 36855.30 ( 3.44%) Hmean unlink1-processes-110 36115.85 ( 0.00%) 39843.91 ( 10.32%) Hmean unlink1-processes-141 32546.96 ( 0.00%) 35418.52 ( 8.82%) Hmean unlink1-processes-172 34674.79 ( 0.00%) 36899.21 ( 6.42%) Hmean unlink1-processes-203 37303.11 ( 0.00%) 36393.04 ( -2.44%) Hmean unlink1-processes-224 35712.13 ( 0.00%) 36685.96 ( 2.73%) == 2. ppc64le == Avg runtime set_task_state(): 938 msecs Avg runtime set_current_state: 940 msecs vanilla dirty Hmean unlink1-processes-2 19269.19 ( 0.00%) 30704.50 ( 59.35%) Hmean unlink1-processes-5 20106.15 ( 0.00%) 21804.15 ( 8.45%) Hmean unlink1-processes-8 17496.97 ( 0.00%) 17243.28 ( -1.45%) Hmean unlink1-processes-12 14224.15 ( 0.00%) 17240.21 ( 21.20%) Hmean unlink1-processes-21 14155.66 ( 0.00%) 15681.23 ( 10.78%) Hmean unlink1-processes-30 14450.70 ( 0.00%) 15995.83 ( 10.69%) Hmean unlink1-processes-48 16945.57 ( 0.00%) 16370.42 ( -3.39%) Hmean unlink1-processes-79 15788.39 ( 0.00%) 14639.27 ( -7.28%) Hmean unlink1-processes-110 14268.48 ( 0.00%) 14377.40 ( 0.76%) Hmean unlink1-processes-141 14023.65 ( 0.00%) 16271.69 ( 16.03%) Hmean unlink1-processes-172 13417.62 ( 0.00%) 16067.55 ( 19.75%) Hmean unlink1-processes-203 15293.08 ( 0.00%) 15440.40 ( 0.96%) Hmean unlink1-processes-234 13719.32 ( 0.00%) 16190.74 ( 18.01%) Hmean unlink1-processes-265 16400.97 ( 0.00%) 16115.22 ( -1.74%) Hmean unlink1-processes-296 14388.60 ( 0.00%) 16216.13 ( 12.70%) Hmean unlink1-processes-320 15771.85 ( 0.00%) 15905.96 ( 0.85%) x86-64 (known to be fast for get_current()/this_cpu_read_stable() caching) and ppc64 (with paca) show similar improvements in the unlink microbenches. The small delta for ppc64 (2ms), does not represent the gains on the unlink runs. In the case of x86, there was a decent amount of variation in the latency runs, but always within a 20 to 50ms increase), ppc was more constant. Signed-off-by: Davidlohr Bueso <dbueso@suse.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dave@stgolabs.net Cc: mark.rutland@arm.com Link: http://lkml.kernel.org/r/1483479794-14013-5-git-send-email-dave@stgolabs.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-01-04 05:43:14 +08:00
set_current_state(state);
for (;;) {
bool first;
/*
* Once we hold wait_lock, we're serialized against
* mutex_unlock() handing the lock off to us, do a trylock
* before testing the error conditions to make sure we pick up
* the handoff.
*/
if (__mutex_trylock(lock))
goto acquired;
/*
* Check for signals and kill conditions while holding
* wait_lock. This ensures the lock cancellation is ordered
* against mutex_unlock() and wake-ups do not go missing.
*/
if (signal_pending_state(state, current)) {
ret = -EINTR;
goto err;
}
if (ww_ctx) {
ret = __ww_mutex_check_kill(lock, &waiter, ww_ctx);
if (ret)
goto err;
}
raw_spin_unlock(&lock->wait_lock);
schedule_preempt_disabled();
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
first = __mutex_waiter_is_first(lock, &waiter);
sched/core: Remove set_task_state() This is a nasty interface and setting the state of a foreign task must not be done. As of the following commit: be628be0956 ("bcache: Make gc wakeup sane, remove set_task_state()") ... everyone in the kernel calls set_task_state() with current, allowing the helper to be removed. However, as the comment indicates, it is still around for those archs where computing current is more expensive than using a pointer, at least in theory. An important arch that is affected is arm64, however this has been addressed now [1] and performance is up to par making no difference with either calls. Of all the callers, if any, it's the locking bits that would care most about this -- ie: we end up passing a tsk pointer to a lot of the lock slowpath, and setting ->state on that. The following numbers are based on two tests: a custom ad-hoc microbenchmark that just measures latencies (for ~65 million calls) between get_task_state() vs get_current_state(). Secondly for a higher overview, an unlink microbenchmark was used, which pounds on a single file with open, close,unlink combos with increasing thread counts (up to 4x ncpus). While the workload is quite unrealistic, it does contend a lot on the inode mutex or now rwsem. [1] https://lkml.kernel.org/r/1483468021-8237-1-git-send-email-mark.rutland@arm.com == 1. x86-64 == Avg runtime set_task_state(): 601 msecs Avg runtime set_current_state(): 552 msecs vanilla dirty Hmean unlink1-processes-2 36089.26 ( 0.00%) 38977.33 ( 8.00%) Hmean unlink1-processes-5 28555.01 ( 0.00%) 29832.55 ( 4.28%) Hmean unlink1-processes-8 37323.75 ( 0.00%) 44974.57 ( 20.50%) Hmean unlink1-processes-12 43571.88 ( 0.00%) 44283.01 ( 1.63%) Hmean unlink1-processes-21 34431.52 ( 0.00%) 38284.45 ( 11.19%) Hmean unlink1-processes-30 34813.26 ( 0.00%) 37975.17 ( 9.08%) Hmean unlink1-processes-48 37048.90 ( 0.00%) 39862.78 ( 7.59%) Hmean unlink1-processes-79 35630.01 ( 0.00%) 36855.30 ( 3.44%) Hmean unlink1-processes-110 36115.85 ( 0.00%) 39843.91 ( 10.32%) Hmean unlink1-processes-141 32546.96 ( 0.00%) 35418.52 ( 8.82%) Hmean unlink1-processes-172 34674.79 ( 0.00%) 36899.21 ( 6.42%) Hmean unlink1-processes-203 37303.11 ( 0.00%) 36393.04 ( -2.44%) Hmean unlink1-processes-224 35712.13 ( 0.00%) 36685.96 ( 2.73%) == 2. ppc64le == Avg runtime set_task_state(): 938 msecs Avg runtime set_current_state: 940 msecs vanilla dirty Hmean unlink1-processes-2 19269.19 ( 0.00%) 30704.50 ( 59.35%) Hmean unlink1-processes-5 20106.15 ( 0.00%) 21804.15 ( 8.45%) Hmean unlink1-processes-8 17496.97 ( 0.00%) 17243.28 ( -1.45%) Hmean unlink1-processes-12 14224.15 ( 0.00%) 17240.21 ( 21.20%) Hmean unlink1-processes-21 14155.66 ( 0.00%) 15681.23 ( 10.78%) Hmean unlink1-processes-30 14450.70 ( 0.00%) 15995.83 ( 10.69%) Hmean unlink1-processes-48 16945.57 ( 0.00%) 16370.42 ( -3.39%) Hmean unlink1-processes-79 15788.39 ( 0.00%) 14639.27 ( -7.28%) Hmean unlink1-processes-110 14268.48 ( 0.00%) 14377.40 ( 0.76%) Hmean unlink1-processes-141 14023.65 ( 0.00%) 16271.69 ( 16.03%) Hmean unlink1-processes-172 13417.62 ( 0.00%) 16067.55 ( 19.75%) Hmean unlink1-processes-203 15293.08 ( 0.00%) 15440.40 ( 0.96%) Hmean unlink1-processes-234 13719.32 ( 0.00%) 16190.74 ( 18.01%) Hmean unlink1-processes-265 16400.97 ( 0.00%) 16115.22 ( -1.74%) Hmean unlink1-processes-296 14388.60 ( 0.00%) 16216.13 ( 12.70%) Hmean unlink1-processes-320 15771.85 ( 0.00%) 15905.96 ( 0.85%) x86-64 (known to be fast for get_current()/this_cpu_read_stable() caching) and ppc64 (with paca) show similar improvements in the unlink microbenches. The small delta for ppc64 (2ms), does not represent the gains on the unlink runs. In the case of x86, there was a decent amount of variation in the latency runs, but always within a 20 to 50ms increase), ppc was more constant. Signed-off-by: Davidlohr Bueso <dbueso@suse.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dave@stgolabs.net Cc: mark.rutland@arm.com Link: http://lkml.kernel.org/r/1483479794-14013-5-git-send-email-dave@stgolabs.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-01-04 05:43:14 +08:00
set_current_state(state);
/*
* Here we order against unlock; we must either see it change
* state back to RUNNING and fall through the next schedule(),
* or we must see its unlock and acquire.
*/
if (__mutex_trylock_or_handoff(lock, first) ||
(first && mutex_optimistic_spin(lock, ww_ctx, &waiter)))
break;
raw_spin_lock(&lock->wait_lock);
}
raw_spin_lock(&lock->wait_lock);
acquired:
sched/core: Remove set_task_state() This is a nasty interface and setting the state of a foreign task must not be done. As of the following commit: be628be0956 ("bcache: Make gc wakeup sane, remove set_task_state()") ... everyone in the kernel calls set_task_state() with current, allowing the helper to be removed. However, as the comment indicates, it is still around for those archs where computing current is more expensive than using a pointer, at least in theory. An important arch that is affected is arm64, however this has been addressed now [1] and performance is up to par making no difference with either calls. Of all the callers, if any, it's the locking bits that would care most about this -- ie: we end up passing a tsk pointer to a lot of the lock slowpath, and setting ->state on that. The following numbers are based on two tests: a custom ad-hoc microbenchmark that just measures latencies (for ~65 million calls) between get_task_state() vs get_current_state(). Secondly for a higher overview, an unlink microbenchmark was used, which pounds on a single file with open, close,unlink combos with increasing thread counts (up to 4x ncpus). While the workload is quite unrealistic, it does contend a lot on the inode mutex or now rwsem. [1] https://lkml.kernel.org/r/1483468021-8237-1-git-send-email-mark.rutland@arm.com == 1. x86-64 == Avg runtime set_task_state(): 601 msecs Avg runtime set_current_state(): 552 msecs vanilla dirty Hmean unlink1-processes-2 36089.26 ( 0.00%) 38977.33 ( 8.00%) Hmean unlink1-processes-5 28555.01 ( 0.00%) 29832.55 ( 4.28%) Hmean unlink1-processes-8 37323.75 ( 0.00%) 44974.57 ( 20.50%) Hmean unlink1-processes-12 43571.88 ( 0.00%) 44283.01 ( 1.63%) Hmean unlink1-processes-21 34431.52 ( 0.00%) 38284.45 ( 11.19%) Hmean unlink1-processes-30 34813.26 ( 0.00%) 37975.17 ( 9.08%) Hmean unlink1-processes-48 37048.90 ( 0.00%) 39862.78 ( 7.59%) Hmean unlink1-processes-79 35630.01 ( 0.00%) 36855.30 ( 3.44%) Hmean unlink1-processes-110 36115.85 ( 0.00%) 39843.91 ( 10.32%) Hmean unlink1-processes-141 32546.96 ( 0.00%) 35418.52 ( 8.82%) Hmean unlink1-processes-172 34674.79 ( 0.00%) 36899.21 ( 6.42%) Hmean unlink1-processes-203 37303.11 ( 0.00%) 36393.04 ( -2.44%) Hmean unlink1-processes-224 35712.13 ( 0.00%) 36685.96 ( 2.73%) == 2. ppc64le == Avg runtime set_task_state(): 938 msecs Avg runtime set_current_state: 940 msecs vanilla dirty Hmean unlink1-processes-2 19269.19 ( 0.00%) 30704.50 ( 59.35%) Hmean unlink1-processes-5 20106.15 ( 0.00%) 21804.15 ( 8.45%) Hmean unlink1-processes-8 17496.97 ( 0.00%) 17243.28 ( -1.45%) Hmean unlink1-processes-12 14224.15 ( 0.00%) 17240.21 ( 21.20%) Hmean unlink1-processes-21 14155.66 ( 0.00%) 15681.23 ( 10.78%) Hmean unlink1-processes-30 14450.70 ( 0.00%) 15995.83 ( 10.69%) Hmean unlink1-processes-48 16945.57 ( 0.00%) 16370.42 ( -3.39%) Hmean unlink1-processes-79 15788.39 ( 0.00%) 14639.27 ( -7.28%) Hmean unlink1-processes-110 14268.48 ( 0.00%) 14377.40 ( 0.76%) Hmean unlink1-processes-141 14023.65 ( 0.00%) 16271.69 ( 16.03%) Hmean unlink1-processes-172 13417.62 ( 0.00%) 16067.55 ( 19.75%) Hmean unlink1-processes-203 15293.08 ( 0.00%) 15440.40 ( 0.96%) Hmean unlink1-processes-234 13719.32 ( 0.00%) 16190.74 ( 18.01%) Hmean unlink1-processes-265 16400.97 ( 0.00%) 16115.22 ( -1.74%) Hmean unlink1-processes-296 14388.60 ( 0.00%) 16216.13 ( 12.70%) Hmean unlink1-processes-320 15771.85 ( 0.00%) 15905.96 ( 0.85%) x86-64 (known to be fast for get_current()/this_cpu_read_stable() caching) and ppc64 (with paca) show similar improvements in the unlink microbenches. The small delta for ppc64 (2ms), does not represent the gains on the unlink runs. In the case of x86, there was a decent amount of variation in the latency runs, but always within a 20 to 50ms increase), ppc was more constant. Signed-off-by: Davidlohr Bueso <dbueso@suse.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dave@stgolabs.net Cc: mark.rutland@arm.com Link: http://lkml.kernel.org/r/1483479794-14013-5-git-send-email-dave@stgolabs.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-01-04 05:43:14 +08:00
__set_current_state(TASK_RUNNING);
if (ww_ctx) {
locking: Implement an algorithm choice for Wound-Wait mutexes The current Wound-Wait mutex algorithm is actually not Wound-Wait but Wait-Die. Implement also Wound-Wait as a per-ww-class choice. Wound-Wait is, contrary to Wait-Die a preemptive algorithm and is known to generate fewer backoffs. Testing reveals that this is true if the number of simultaneous contending transactions is small. As the number of simultaneous contending threads increases, Wait-Wound becomes inferior to Wait-Die in terms of elapsed time. Possibly due to the larger number of held locks of sleeping transactions. Update documentation and callers. Timings using git://people.freedesktop.org/~thomash/ww_mutex_test tag patch-18-06-15 Each thread runs 100000 batches of lock / unlock 800 ww mutexes randomly chosen out of 100000. Four core Intel x86_64: Algorithm #threads Rollbacks time Wound-Wait 4 ~100 ~17s. Wait-Die 4 ~150000 ~19s. Wound-Wait 16 ~360000 ~109s. Wait-Die 16 ~450000 ~82s. Cc: Ingo Molnar <mingo@redhat.com> Cc: Jonathan Corbet <corbet@lwn.net> Cc: Gustavo Padovan <gustavo@padovan.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Sean Paul <seanpaul@chromium.org> Cc: David Airlie <airlied@linux.ie> Cc: Davidlohr Bueso <dave@stgolabs.net> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Kate Stewart <kstewart@linuxfoundation.org> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: linux-doc@vger.kernel.org Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Co-authored-by: Peter Zijlstra <peterz@infradead.org> Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Ingo Molnar <mingo@kernel.org>
2018-06-15 16:17:38 +08:00
/*
* Wound-Wait; we stole the lock (!first_waiter), check the
* waiters as anyone might want to wound us.
*/
if (!ww_ctx->is_wait_die &&
!__mutex_waiter_is_first(lock, &waiter))
__ww_mutex_check_waiters(lock, ww_ctx);
}
__mutex_remove_waiter(lock, &waiter);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
mutex: Do not unnecessarily deal with waiters Upon entering the slowpath, we immediately attempt to acquire the lock by checking if it is already unlocked. If we are lucky enough that this is the case, then we don't need to deal with any waiter related logic. Furthermore any checks for an empty wait_list are unnecessary as we already know that count is non-negative and hence no one is waiting for the lock. Move the count check and xchg calls to be done before any waiters are setup - including waiter debugging. Upon failure to acquire the lock, the xchg sets the counter to 0, instead of -1 as it was originally. This can be done here since we set it back to -1 right at the beginning of the loop so other waiters are woken up when the lock is released. When tested on a 8-socket (80 core) system against a vanilla 3.10-rc1 kernel, this patch provides some small performance benefits (+2-6%). While it could be considered in the noise level, the average percentages were stable across multiple runs and no performance regressions were seen. Two big winners, for small amounts of users (10-100), were the short and compute workloads had a +19.36% and +%15.76% in jobs per minute. Also change some break statements to 'goto slowpath', which IMO makes a little more intuitive to read. Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Acked-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1372450398.2106.1.camel@buesod1.americas.hpqcorp.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-29 04:13:18 +08:00
debug_mutex_free_waiter(&waiter);
mutex: Do not unnecessarily deal with waiters Upon entering the slowpath, we immediately attempt to acquire the lock by checking if it is already unlocked. If we are lucky enough that this is the case, then we don't need to deal with any waiter related logic. Furthermore any checks for an empty wait_list are unnecessary as we already know that count is non-negative and hence no one is waiting for the lock. Move the count check and xchg calls to be done before any waiters are setup - including waiter debugging. Upon failure to acquire the lock, the xchg sets the counter to 0, instead of -1 as it was originally. This can be done here since we set it back to -1 right at the beginning of the loop so other waiters are woken up when the lock is released. When tested on a 8-socket (80 core) system against a vanilla 3.10-rc1 kernel, this patch provides some small performance benefits (+2-6%). While it could be considered in the noise level, the average percentages were stable across multiple runs and no performance regressions were seen. Two big winners, for small amounts of users (10-100), were the short and compute workloads had a +19.36% and +%15.76% in jobs per minute. Also change some break statements to 'goto slowpath', which IMO makes a little more intuitive to read. Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com> Acked-by: Rik van Riel <riel@redhat.com> Acked-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1372450398.2106.1.camel@buesod1.americas.hpqcorp.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-29 04:13:18 +08:00
skip_wait:
/* got the lock - cleanup and rejoice! */
lock_acquired(&lock->dep_map, ip);
if (ww_ctx)
ww_mutex_lock_acquired(ww, ww_ctx);
raw_spin_unlock(&lock->wait_lock);
preempt_enable();
return 0;
err:
sched/core: Remove set_task_state() This is a nasty interface and setting the state of a foreign task must not be done. As of the following commit: be628be0956 ("bcache: Make gc wakeup sane, remove set_task_state()") ... everyone in the kernel calls set_task_state() with current, allowing the helper to be removed. However, as the comment indicates, it is still around for those archs where computing current is more expensive than using a pointer, at least in theory. An important arch that is affected is arm64, however this has been addressed now [1] and performance is up to par making no difference with either calls. Of all the callers, if any, it's the locking bits that would care most about this -- ie: we end up passing a tsk pointer to a lot of the lock slowpath, and setting ->state on that. The following numbers are based on two tests: a custom ad-hoc microbenchmark that just measures latencies (for ~65 million calls) between get_task_state() vs get_current_state(). Secondly for a higher overview, an unlink microbenchmark was used, which pounds on a single file with open, close,unlink combos with increasing thread counts (up to 4x ncpus). While the workload is quite unrealistic, it does contend a lot on the inode mutex or now rwsem. [1] https://lkml.kernel.org/r/1483468021-8237-1-git-send-email-mark.rutland@arm.com == 1. x86-64 == Avg runtime set_task_state(): 601 msecs Avg runtime set_current_state(): 552 msecs vanilla dirty Hmean unlink1-processes-2 36089.26 ( 0.00%) 38977.33 ( 8.00%) Hmean unlink1-processes-5 28555.01 ( 0.00%) 29832.55 ( 4.28%) Hmean unlink1-processes-8 37323.75 ( 0.00%) 44974.57 ( 20.50%) Hmean unlink1-processes-12 43571.88 ( 0.00%) 44283.01 ( 1.63%) Hmean unlink1-processes-21 34431.52 ( 0.00%) 38284.45 ( 11.19%) Hmean unlink1-processes-30 34813.26 ( 0.00%) 37975.17 ( 9.08%) Hmean unlink1-processes-48 37048.90 ( 0.00%) 39862.78 ( 7.59%) Hmean unlink1-processes-79 35630.01 ( 0.00%) 36855.30 ( 3.44%) Hmean unlink1-processes-110 36115.85 ( 0.00%) 39843.91 ( 10.32%) Hmean unlink1-processes-141 32546.96 ( 0.00%) 35418.52 ( 8.82%) Hmean unlink1-processes-172 34674.79 ( 0.00%) 36899.21 ( 6.42%) Hmean unlink1-processes-203 37303.11 ( 0.00%) 36393.04 ( -2.44%) Hmean unlink1-processes-224 35712.13 ( 0.00%) 36685.96 ( 2.73%) == 2. ppc64le == Avg runtime set_task_state(): 938 msecs Avg runtime set_current_state: 940 msecs vanilla dirty Hmean unlink1-processes-2 19269.19 ( 0.00%) 30704.50 ( 59.35%) Hmean unlink1-processes-5 20106.15 ( 0.00%) 21804.15 ( 8.45%) Hmean unlink1-processes-8 17496.97 ( 0.00%) 17243.28 ( -1.45%) Hmean unlink1-processes-12 14224.15 ( 0.00%) 17240.21 ( 21.20%) Hmean unlink1-processes-21 14155.66 ( 0.00%) 15681.23 ( 10.78%) Hmean unlink1-processes-30 14450.70 ( 0.00%) 15995.83 ( 10.69%) Hmean unlink1-processes-48 16945.57 ( 0.00%) 16370.42 ( -3.39%) Hmean unlink1-processes-79 15788.39 ( 0.00%) 14639.27 ( -7.28%) Hmean unlink1-processes-110 14268.48 ( 0.00%) 14377.40 ( 0.76%) Hmean unlink1-processes-141 14023.65 ( 0.00%) 16271.69 ( 16.03%) Hmean unlink1-processes-172 13417.62 ( 0.00%) 16067.55 ( 19.75%) Hmean unlink1-processes-203 15293.08 ( 0.00%) 15440.40 ( 0.96%) Hmean unlink1-processes-234 13719.32 ( 0.00%) 16190.74 ( 18.01%) Hmean unlink1-processes-265 16400.97 ( 0.00%) 16115.22 ( -1.74%) Hmean unlink1-processes-296 14388.60 ( 0.00%) 16216.13 ( 12.70%) Hmean unlink1-processes-320 15771.85 ( 0.00%) 15905.96 ( 0.85%) x86-64 (known to be fast for get_current()/this_cpu_read_stable() caching) and ppc64 (with paca) show similar improvements in the unlink microbenches. The small delta for ppc64 (2ms), does not represent the gains on the unlink runs. In the case of x86, there was a decent amount of variation in the latency runs, but always within a 20 to 50ms increase), ppc was more constant. Signed-off-by: Davidlohr Bueso <dbueso@suse.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: dave@stgolabs.net Cc: mark.rutland@arm.com Link: http://lkml.kernel.org/r/1483479794-14013-5-git-send-email-dave@stgolabs.net Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-01-04 05:43:14 +08:00
__set_current_state(TASK_RUNNING);
__mutex_remove_waiter(lock, &waiter);
err_early_kill:
raw_spin_unlock(&lock->wait_lock);
debug_mutex_free_waiter(&waiter);
2019-09-20 00:09:40 +08:00
mutex_release(&lock->dep_map, ip);
preempt_enable();
return ret;
}
static int __sched
__mutex_lock(struct mutex *lock, unsigned int state, unsigned int subclass,
struct lockdep_map *nest_lock, unsigned long ip)
{
return __mutex_lock_common(lock, state, subclass, nest_lock, ip, NULL, false);
}
static int __sched
__ww_mutex_lock(struct mutex *lock, unsigned int state, unsigned int subclass,
unsigned long ip, struct ww_acquire_ctx *ww_ctx)
{
return __mutex_lock_common(lock, state, subclass, NULL, ip, ww_ctx, true);
}
#ifdef CONFIG_DEBUG_LOCK_ALLOC
void __sched
mutex_lock_nested(struct mutex *lock, unsigned int subclass)
{
__mutex_lock(lock, TASK_UNINTERRUPTIBLE, subclass, NULL, _RET_IP_);
}
EXPORT_SYMBOL_GPL(mutex_lock_nested);
void __sched
_mutex_lock_nest_lock(struct mutex *lock, struct lockdep_map *nest)
{
__mutex_lock(lock, TASK_UNINTERRUPTIBLE, 0, nest, _RET_IP_);
}
EXPORT_SYMBOL_GPL(_mutex_lock_nest_lock);
int __sched
mutex_lock_killable_nested(struct mutex *lock, unsigned int subclass)
{
return __mutex_lock(lock, TASK_KILLABLE, subclass, NULL, _RET_IP_);
}
EXPORT_SYMBOL_GPL(mutex_lock_killable_nested);
int __sched
mutex_lock_interruptible_nested(struct mutex *lock, unsigned int subclass)
{
return __mutex_lock(lock, TASK_INTERRUPTIBLE, subclass, NULL, _RET_IP_);
}
EXPORT_SYMBOL_GPL(mutex_lock_interruptible_nested);
void __sched
mutex_lock_io_nested(struct mutex *lock, unsigned int subclass)
{
int token;
might_sleep();
token = io_schedule_prepare();
__mutex_lock_common(lock, TASK_UNINTERRUPTIBLE,
subclass, NULL, _RET_IP_, NULL, 0);
io_schedule_finish(token);
}
EXPORT_SYMBOL_GPL(mutex_lock_io_nested);
mutex: Add w/w mutex slowpath debugging Injects EDEADLK conditions at pseudo-random interval, with exponential backoff up to UINT_MAX (to ensure that every lock operation still completes in a reasonable time). This way we can test the wound slowpath even for ww mutex users where contention is never expected, and the ww deadlock avoidance algorithm is only needed for correctness against malicious userspace. An example would be protecting kernel modesetting properties, which thanks to single-threaded X isn't really expected to contend, ever. I've looked into using the CONFIG_FAULT_INJECTION infrastructure, but decided against it for two reasons: - EDEADLK handling is mandatory for ww mutex users and should never affect the outcome of a syscall. This is in contrast to -ENOMEM injection. So fine configurability isn't required. - The fault injection framework only allows to set a simple probability for failure. Now the probability that a ww mutex acquire stage with N locks will never complete (due to too many injected EDEADLK backoffs) is zero. But the expected number of ww_mutex_lock operations for the completely uncontended case would be O(exp(N)). The per-acuiqire ctx exponential backoff solution choosen here only results in O(log N) overhead due to injection and so O(log N * N) lock operations. This way we can fail with high probability (and so have good test coverage even for fancy backoff and lock acquisition paths) without running into patalogical cases. Note that EDEADLK will only ever be injected when we managed to acquire the lock. This prevents any behaviour changes for users which rely on the EALREADY semantics. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: rostedt@goodmis.org Cc: daniel@ffwll.ch Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20130620113117.4001.21681.stgit@patser Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-20 19:31:17 +08:00
static inline int
ww_mutex_deadlock_injection(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
#ifdef CONFIG_DEBUG_WW_MUTEX_SLOWPATH
unsigned tmp;
if (ctx->deadlock_inject_countdown-- == 0) {
tmp = ctx->deadlock_inject_interval;
if (tmp > UINT_MAX/4)
tmp = UINT_MAX;
else
tmp = tmp*2 + tmp + tmp/2;
ctx->deadlock_inject_interval = tmp;
ctx->deadlock_inject_countdown = tmp;
ctx->contending_lock = lock;
ww_mutex_unlock(lock);
return -EDEADLK;
}
#endif
return 0;
}
int __sched
ww_mutex_lock(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
mutex: Add w/w mutex slowpath debugging Injects EDEADLK conditions at pseudo-random interval, with exponential backoff up to UINT_MAX (to ensure that every lock operation still completes in a reasonable time). This way we can test the wound slowpath even for ww mutex users where contention is never expected, and the ww deadlock avoidance algorithm is only needed for correctness against malicious userspace. An example would be protecting kernel modesetting properties, which thanks to single-threaded X isn't really expected to contend, ever. I've looked into using the CONFIG_FAULT_INJECTION infrastructure, but decided against it for two reasons: - EDEADLK handling is mandatory for ww mutex users and should never affect the outcome of a syscall. This is in contrast to -ENOMEM injection. So fine configurability isn't required. - The fault injection framework only allows to set a simple probability for failure. Now the probability that a ww mutex acquire stage with N locks will never complete (due to too many injected EDEADLK backoffs) is zero. But the expected number of ww_mutex_lock operations for the completely uncontended case would be O(exp(N)). The per-acuiqire ctx exponential backoff solution choosen here only results in O(log N) overhead due to injection and so O(log N * N) lock operations. This way we can fail with high probability (and so have good test coverage even for fancy backoff and lock acquisition paths) without running into patalogical cases. Note that EDEADLK will only ever be injected when we managed to acquire the lock. This prevents any behaviour changes for users which rely on the EALREADY semantics. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: rostedt@goodmis.org Cc: daniel@ffwll.ch Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20130620113117.4001.21681.stgit@patser Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-20 19:31:17 +08:00
int ret;
might_sleep();
ret = __ww_mutex_lock(&lock->base, TASK_UNINTERRUPTIBLE,
0, _RET_IP_, ctx);
if (!ret && ctx && ctx->acquired > 1)
mutex: Add w/w mutex slowpath debugging Injects EDEADLK conditions at pseudo-random interval, with exponential backoff up to UINT_MAX (to ensure that every lock operation still completes in a reasonable time). This way we can test the wound slowpath even for ww mutex users where contention is never expected, and the ww deadlock avoidance algorithm is only needed for correctness against malicious userspace. An example would be protecting kernel modesetting properties, which thanks to single-threaded X isn't really expected to contend, ever. I've looked into using the CONFIG_FAULT_INJECTION infrastructure, but decided against it for two reasons: - EDEADLK handling is mandatory for ww mutex users and should never affect the outcome of a syscall. This is in contrast to -ENOMEM injection. So fine configurability isn't required. - The fault injection framework only allows to set a simple probability for failure. Now the probability that a ww mutex acquire stage with N locks will never complete (due to too many injected EDEADLK backoffs) is zero. But the expected number of ww_mutex_lock operations for the completely uncontended case would be O(exp(N)). The per-acuiqire ctx exponential backoff solution choosen here only results in O(log N) overhead due to injection and so O(log N * N) lock operations. This way we can fail with high probability (and so have good test coverage even for fancy backoff and lock acquisition paths) without running into patalogical cases. Note that EDEADLK will only ever be injected when we managed to acquire the lock. This prevents any behaviour changes for users which rely on the EALREADY semantics. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: rostedt@goodmis.org Cc: daniel@ffwll.ch Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20130620113117.4001.21681.stgit@patser Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-20 19:31:17 +08:00
return ww_mutex_deadlock_injection(lock, ctx);
return ret;
}
EXPORT_SYMBOL_GPL(ww_mutex_lock);
int __sched
ww_mutex_lock_interruptible(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
mutex: Add w/w mutex slowpath debugging Injects EDEADLK conditions at pseudo-random interval, with exponential backoff up to UINT_MAX (to ensure that every lock operation still completes in a reasonable time). This way we can test the wound slowpath even for ww mutex users where contention is never expected, and the ww deadlock avoidance algorithm is only needed for correctness against malicious userspace. An example would be protecting kernel modesetting properties, which thanks to single-threaded X isn't really expected to contend, ever. I've looked into using the CONFIG_FAULT_INJECTION infrastructure, but decided against it for two reasons: - EDEADLK handling is mandatory for ww mutex users and should never affect the outcome of a syscall. This is in contrast to -ENOMEM injection. So fine configurability isn't required. - The fault injection framework only allows to set a simple probability for failure. Now the probability that a ww mutex acquire stage with N locks will never complete (due to too many injected EDEADLK backoffs) is zero. But the expected number of ww_mutex_lock operations for the completely uncontended case would be O(exp(N)). The per-acuiqire ctx exponential backoff solution choosen here only results in O(log N) overhead due to injection and so O(log N * N) lock operations. This way we can fail with high probability (and so have good test coverage even for fancy backoff and lock acquisition paths) without running into patalogical cases. Note that EDEADLK will only ever be injected when we managed to acquire the lock. This prevents any behaviour changes for users which rely on the EALREADY semantics. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: rostedt@goodmis.org Cc: daniel@ffwll.ch Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20130620113117.4001.21681.stgit@patser Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-20 19:31:17 +08:00
int ret;
might_sleep();
ret = __ww_mutex_lock(&lock->base, TASK_INTERRUPTIBLE,
0, _RET_IP_, ctx);
mutex: Add w/w mutex slowpath debugging Injects EDEADLK conditions at pseudo-random interval, with exponential backoff up to UINT_MAX (to ensure that every lock operation still completes in a reasonable time). This way we can test the wound slowpath even for ww mutex users where contention is never expected, and the ww deadlock avoidance algorithm is only needed for correctness against malicious userspace. An example would be protecting kernel modesetting properties, which thanks to single-threaded X isn't really expected to contend, ever. I've looked into using the CONFIG_FAULT_INJECTION infrastructure, but decided against it for two reasons: - EDEADLK handling is mandatory for ww mutex users and should never affect the outcome of a syscall. This is in contrast to -ENOMEM injection. So fine configurability isn't required. - The fault injection framework only allows to set a simple probability for failure. Now the probability that a ww mutex acquire stage with N locks will never complete (due to too many injected EDEADLK backoffs) is zero. But the expected number of ww_mutex_lock operations for the completely uncontended case would be O(exp(N)). The per-acuiqire ctx exponential backoff solution choosen here only results in O(log N) overhead due to injection and so O(log N * N) lock operations. This way we can fail with high probability (and so have good test coverage even for fancy backoff and lock acquisition paths) without running into patalogical cases. Note that EDEADLK will only ever be injected when we managed to acquire the lock. This prevents any behaviour changes for users which rely on the EALREADY semantics. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: rostedt@goodmis.org Cc: daniel@ffwll.ch Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20130620113117.4001.21681.stgit@patser Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-20 19:31:17 +08:00
if (!ret && ctx && ctx->acquired > 1)
mutex: Add w/w mutex slowpath debugging Injects EDEADLK conditions at pseudo-random interval, with exponential backoff up to UINT_MAX (to ensure that every lock operation still completes in a reasonable time). This way we can test the wound slowpath even for ww mutex users where contention is never expected, and the ww deadlock avoidance algorithm is only needed for correctness against malicious userspace. An example would be protecting kernel modesetting properties, which thanks to single-threaded X isn't really expected to contend, ever. I've looked into using the CONFIG_FAULT_INJECTION infrastructure, but decided against it for two reasons: - EDEADLK handling is mandatory for ww mutex users and should never affect the outcome of a syscall. This is in contrast to -ENOMEM injection. So fine configurability isn't required. - The fault injection framework only allows to set a simple probability for failure. Now the probability that a ww mutex acquire stage with N locks will never complete (due to too many injected EDEADLK backoffs) is zero. But the expected number of ww_mutex_lock operations for the completely uncontended case would be O(exp(N)). The per-acuiqire ctx exponential backoff solution choosen here only results in O(log N) overhead due to injection and so O(log N * N) lock operations. This way we can fail with high probability (and so have good test coverage even for fancy backoff and lock acquisition paths) without running into patalogical cases. Note that EDEADLK will only ever be injected when we managed to acquire the lock. This prevents any behaviour changes for users which rely on the EALREADY semantics. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: rostedt@goodmis.org Cc: daniel@ffwll.ch Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20130620113117.4001.21681.stgit@patser Signed-off-by: Ingo Molnar <mingo@kernel.org>
2013-06-20 19:31:17 +08:00
return ww_mutex_deadlock_injection(lock, ctx);
return ret;
}
EXPORT_SYMBOL_GPL(ww_mutex_lock_interruptible);
#endif
/*
* Release the lock, slowpath:
*/
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
static noinline void __sched __mutex_unlock_slowpath(struct mutex *lock, unsigned long ip)
{
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
struct task_struct *next = NULL;
DEFINE_WAKE_Q(wake_q);
unsigned long owner;
2019-09-20 00:09:40 +08:00
mutex_release(&lock->dep_map, ip);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
/*
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
* Release the lock before (potentially) taking the spinlock such that
* other contenders can get on with things ASAP.
*
* Except when HANDOFF, in that case we must not clear the owner field,
* but instead set it to the top waiter.
*/
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
owner = atomic_long_read(&lock->owner);
for (;;) {
MUTEX_WARN_ON(__owner_task(owner) != current);
MUTEX_WARN_ON(owner & MUTEX_FLAG_PICKUP);
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
if (owner & MUTEX_FLAG_HANDOFF)
break;
if (atomic_long_try_cmpxchg_release(&lock->owner, &owner, __owner_flags(owner))) {
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
if (owner & MUTEX_FLAG_WAITERS)
break;
return;
}
}
raw_spin_lock(&lock->wait_lock);
debug_mutex_unlock(lock);
if (!list_empty(&lock->wait_list)) {
/* get the first entry from the wait-list: */
struct mutex_waiter *waiter =
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
list_first_entry(&lock->wait_list,
struct mutex_waiter, list);
next = waiter->task;
debug_mutex_wake_waiter(lock, waiter);
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
wake_q_add(&wake_q, next);
}
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
if (owner & MUTEX_FLAG_HANDOFF)
__mutex_handoff(lock, next);
raw_spin_unlock(&lock->wait_lock);
locking/mutex: Add lock handoff to avoid starvation Implement lock handoff to avoid lock starvation. Lock starvation is possible because mutex_lock() allows lock stealing, where a running (or optimistic spinning) task beats the woken waiter to the acquire. Lock stealing is an important performance optimization because waiting for a waiter to wake up and get runtime can take a significant time, during which everyboy would stall on the lock. The down-side is of course that it allows for starvation. This patch has the waiter requesting a handoff if it fails to acquire the lock upon waking. This re-introduces some of the wait time, because once we do a handoff we have to wait for the waiter to wake up again. A future patch will add a round of optimistic spinning to attempt to alleviate this penalty, but if that turns out to not be enough, we can add a counter and only request handoff after multiple failed wakeups. There are a few tricky implementation details: - accepting a handoff must only be done in the wait-loop. Since the handoff condition is owner == current, it can easily cause recursive locking trouble. - accepting the handoff must be careful to provide the ACQUIRE semantics. - having the HANDOFF bit set on unlock requires care, we must not clear the owner. - we must be careful to not leave HANDOFF set after we've acquired the lock. The tricky scenario is setting the HANDOFF bit on an unlocked mutex. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Waiman Long <Waiman.Long@hpe.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 20:40:16 +08:00
wake_up_q(&wake_q);
}
#ifndef CONFIG_DEBUG_LOCK_ALLOC
/*
* Here come the less common (and hence less performance-critical) APIs:
* mutex_lock_interruptible() and mutex_trylock().
*/
static noinline int __sched
__mutex_lock_killable_slowpath(struct mutex *lock);
static noinline int __sched
__mutex_lock_interruptible_slowpath(struct mutex *lock);
/**
* mutex_lock_interruptible() - Acquire the mutex, interruptible by signals.
* @lock: The mutex to be acquired.
*
* Lock the mutex like mutex_lock(). If a signal is delivered while the
* process is sleeping, this function will return without acquiring the
* mutex.
*
* Context: Process context.
* Return: 0 if the lock was successfully acquired or %-EINTR if a
* signal arrived.
*/
int __sched mutex_lock_interruptible(struct mutex *lock)
{
might_sleep();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
if (__mutex_trylock_fast(lock))
return 0;
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
return __mutex_lock_interruptible_slowpath(lock);
}
EXPORT_SYMBOL(mutex_lock_interruptible);
/**
* mutex_lock_killable() - Acquire the mutex, interruptible by fatal signals.
* @lock: The mutex to be acquired.
*
* Lock the mutex like mutex_lock(). If a signal which will be fatal to
* the current process is delivered while the process is sleeping, this
* function will return without acquiring the mutex.
*
* Context: Process context.
* Return: 0 if the lock was successfully acquired or %-EINTR if a
* fatal signal arrived.
*/
int __sched mutex_lock_killable(struct mutex *lock)
{
might_sleep();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
if (__mutex_trylock_fast(lock))
return 0;
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
return __mutex_lock_killable_slowpath(lock);
}
EXPORT_SYMBOL(mutex_lock_killable);
/**
* mutex_lock_io() - Acquire the mutex and mark the process as waiting for I/O
* @lock: The mutex to be acquired.
*
* Lock the mutex like mutex_lock(). While the task is waiting for this
* mutex, it will be accounted as being in the IO wait state by the
* scheduler.
*
* Context: Process context.
*/
void __sched mutex_lock_io(struct mutex *lock)
{
int token;
token = io_schedule_prepare();
mutex_lock(lock);
io_schedule_finish(token);
}
EXPORT_SYMBOL_GPL(mutex_lock_io);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
static noinline void __sched
__mutex_lock_slowpath(struct mutex *lock)
{
__mutex_lock(lock, TASK_UNINTERRUPTIBLE, 0, NULL, _RET_IP_);
}
static noinline int __sched
__mutex_lock_killable_slowpath(struct mutex *lock)
{
return __mutex_lock(lock, TASK_KILLABLE, 0, NULL, _RET_IP_);
}
static noinline int __sched
__mutex_lock_interruptible_slowpath(struct mutex *lock)
{
return __mutex_lock(lock, TASK_INTERRUPTIBLE, 0, NULL, _RET_IP_);
}
static noinline int __sched
__ww_mutex_lock_slowpath(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
return __ww_mutex_lock(&lock->base, TASK_UNINTERRUPTIBLE, 0,
_RET_IP_, ctx);
}
static noinline int __sched
__ww_mutex_lock_interruptible_slowpath(struct ww_mutex *lock,
struct ww_acquire_ctx *ctx)
{
return __ww_mutex_lock(&lock->base, TASK_INTERRUPTIBLE, 0,
_RET_IP_, ctx);
}
#endif
/**
* mutex_trylock - try to acquire the mutex, without waiting
* @lock: the mutex to be acquired
*
* Try to acquire the mutex atomically. Returns 1 if the mutex
* has been acquired successfully, and 0 on contention.
*
* NOTE: this function follows the spin_trylock() convention, so
* it is negated from the down_trylock() return values! Be careful
* about this when converting semaphore users to mutexes.
*
* This function must not be used in interrupt context. The
* mutex must be released by the same task that acquired it.
*/
int __sched mutex_trylock(struct mutex *lock)
{
bool locked;
MUTEX_WARN_ON(lock->magic != lock);
mutex: implement adaptive spinning Change mutex contention behaviour such that it will sometimes busy wait on acquisition - moving its behaviour closer to that of spinlocks. This concept got ported to mainline from the -rt tree, where it was originally implemented for rtmutexes by Steven Rostedt, based on work by Gregory Haskins. Testing with Ingo's test-mutex application (http://lkml.org/lkml/2006/1/8/50) gave a 345% boost for VFS scalability on my testbox: # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 296604 # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 85870 The key criteria for the busy wait is that the lock owner has to be running on a (different) cpu. The idea is that as long as the owner is running, there is a fair chance it'll release the lock soon, and thus we'll be better off spinning instead of blocking/scheduling. Since regular mutexes (as opposed to rtmutexes) do not atomically track the owner, we add the owner in a non-atomic fashion and deal with the races in the slowpath. Furthermore, to ease the testing of the performance impact of this new code, there is means to disable this behaviour runtime (without having to reboot the system), when scheduler debugging is enabled (CONFIG_SCHED_DEBUG=y), by issuing the following command: # echo NO_OWNER_SPIN > /debug/sched_features This command re-enables spinning again (this is also the default): # echo OWNER_SPIN > /debug/sched_features Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-01-12 21:01:47 +08:00
locked = __mutex_trylock(lock);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
if (locked)
mutex_acquire(&lock->dep_map, 0, 1, _RET_IP_);
mutex: implement adaptive spinning Change mutex contention behaviour such that it will sometimes busy wait on acquisition - moving its behaviour closer to that of spinlocks. This concept got ported to mainline from the -rt tree, where it was originally implemented for rtmutexes by Steven Rostedt, based on work by Gregory Haskins. Testing with Ingo's test-mutex application (http://lkml.org/lkml/2006/1/8/50) gave a 345% boost for VFS scalability on my testbox: # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 296604 # ./test-mutex-shm V 16 10 | grep "^avg ops" avg ops/sec: 85870 The key criteria for the busy wait is that the lock owner has to be running on a (different) cpu. The idea is that as long as the owner is running, there is a fair chance it'll release the lock soon, and thus we'll be better off spinning instead of blocking/scheduling. Since regular mutexes (as opposed to rtmutexes) do not atomically track the owner, we add the owner in a non-atomic fashion and deal with the races in the slowpath. Furthermore, to ease the testing of the performance impact of this new code, there is means to disable this behaviour runtime (without having to reboot the system), when scheduler debugging is enabled (CONFIG_SCHED_DEBUG=y), by issuing the following command: # echo NO_OWNER_SPIN > /debug/sched_features This command re-enables spinning again (this is also the default): # echo OWNER_SPIN > /debug/sched_features Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-01-12 21:01:47 +08:00
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
return locked;
}
EXPORT_SYMBOL(mutex_trylock);
#ifndef CONFIG_DEBUG_LOCK_ALLOC
int __sched
ww_mutex_lock(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
might_sleep();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
if (__mutex_trylock_fast(&lock->base)) {
if (ctx)
ww_mutex_set_context_fastpath(lock, ctx);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
return 0;
}
return __ww_mutex_lock_slowpath(lock, ctx);
}
EXPORT_SYMBOL(ww_mutex_lock);
int __sched
ww_mutex_lock_interruptible(struct ww_mutex *lock, struct ww_acquire_ctx *ctx)
{
might_sleep();
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
if (__mutex_trylock_fast(&lock->base)) {
if (ctx)
ww_mutex_set_context_fastpath(lock, ctx);
locking/mutex: Rework mutex::owner The current mutex implementation has an atomic lock word and a non-atomic owner field. This disparity leads to a number of issues with the current mutex code as it means that we can have a locked mutex without an explicit owner (because the owner field has not been set, or already cleared). This leads to a number of weird corner cases, esp. between the optimistic spinning and debug code. Where the optimistic spinning code needs the owner field updated inside the lock region, the debug code is more relaxed because the whole lock is serialized by the wait_lock. Also, the spinning code itself has a few corner cases where we need to deal with a held lock without an owner field. Furthermore, it becomes even more of a problem when trying to fix starvation cases in the current code. We end up stacking special case on special case. To solve this rework the basic mutex implementation to be a single atomic word that contains the owner and uses the low bits for extra state. This matches how PI futexes and rt_mutex already work. By having the owner an integral part of the lock state a lot of the problems dissapear and we get a better option to deal with starvation cases, direct owner handoff. Changing the basic mutex does however invalidate all the arch specific mutex code; this patch leaves that unused in-place, a later patch will remove that. Tested-by: Jason Low <jason.low2@hpe.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Will Deacon <will.deacon@arm.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-08-23 19:36:04 +08:00
return 0;
}
return __ww_mutex_lock_interruptible_slowpath(lock, ctx);
}
EXPORT_SYMBOL(ww_mutex_lock_interruptible);
#endif
/**
* atomic_dec_and_mutex_lock - return holding mutex if we dec to 0
* @cnt: the atomic which we are to dec
* @lock: the mutex to return holding if we dec to 0
*
* return true and hold lock if we dec to 0, return false otherwise
*/
int atomic_dec_and_mutex_lock(atomic_t *cnt, struct mutex *lock)
{
/* dec if we can't possibly hit 0 */
if (atomic_add_unless(cnt, -1, 1))
return 0;
/* we might hit 0, so take the lock */
mutex_lock(lock);
if (!atomic_dec_and_test(cnt)) {
/* when we actually did the dec, we didn't hit 0 */
mutex_unlock(lock);
return 0;
}
/* we hit 0, and we hold the lock */
return 1;
}
EXPORT_SYMBOL(atomic_dec_and_mutex_lock);